US20010043614A1 - Multi-layer switching apparatus and method - Google Patents
Multi-layer switching apparatus and method Download PDFInfo
- Publication number
- US20010043614A1 US20010043614A1 US09/118,458 US11845898A US2001043614A1 US 20010043614 A1 US20010043614 A1 US 20010043614A1 US 11845898 A US11845898 A US 11845898A US 2001043614 A1 US2001043614 A1 US 2001043614A1
- Authority
- US
- United States
- Prior art keywords
- packet
- memory
- module
- cam
- port
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 47
- 230000015654 memory Effects 0.000 claims description 229
- 239000000872 buffer Substances 0.000 claims description 149
- 230000008569 process Effects 0.000 claims description 39
- 238000012545 processing Methods 0.000 claims description 27
- 230000032683 aging Effects 0.000 claims description 17
- 230000005540 biological transmission Effects 0.000 claims description 13
- 230000006855 networking Effects 0.000 claims description 7
- 238000004891 communication Methods 0.000 claims description 4
- 230000008878 coupling Effects 0.000 claims description 4
- 238000010168 coupling process Methods 0.000 claims description 4
- 238000005859 coupling reaction Methods 0.000 claims description 4
- 238000012546 transfer Methods 0.000 description 56
- RTZKZFJDLAIYFH-UHFFFAOYSA-N Diethyl ether Chemical compound CCOCC RTZKZFJDLAIYFH-UHFFFAOYSA-N 0.000 description 52
- 238000007726 management method Methods 0.000 description 32
- 238000004458 analytical method Methods 0.000 description 26
- 230000006870 function Effects 0.000 description 19
- 239000003795 chemical substances by application Substances 0.000 description 13
- 238000010586 diagram Methods 0.000 description 11
- 230000009471 action Effects 0.000 description 9
- 230000010354 integration Effects 0.000 description 7
- 238000001914 filtration Methods 0.000 description 6
- 230000008859 change Effects 0.000 description 5
- 230000001360 synchronised effect Effects 0.000 description 5
- 238000013461 design Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 4
- 238000012423 maintenance Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 238000012986 modification Methods 0.000 description 4
- 230000011664 signaling Effects 0.000 description 4
- 230000001960 triggered effect Effects 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000001143 conditioned effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 239000004744 fabric Substances 0.000 description 3
- 230000002431 foraging effect Effects 0.000 description 3
- 230000007246 mechanism Effects 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- KKIMDKMETPPURN-UHFFFAOYSA-N 1-(3-(trifluoromethyl)phenyl)piperazine Chemical compound FC(F)(F)C1=CC=CC(N2CCNCC2)=C1 KKIMDKMETPPURN-UHFFFAOYSA-N 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 2
- 230000009977 dual effect Effects 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000001976 improved effect Effects 0.000 description 2
- 230000003993 interaction Effects 0.000 description 2
- 238000012913 prioritisation Methods 0.000 description 2
- 230000008093 supporting effect Effects 0.000 description 2
- 230000002776 aggregation Effects 0.000 description 1
- 238000004220 aggregation Methods 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 230000003139 buffering effect Effects 0.000 description 1
- 238000013329 compounding Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- -1 data path controller Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003116 impacting effect Effects 0.000 description 1
- 230000000977 initiatory effect Effects 0.000 description 1
- 238000009434 installation Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000013468 resource allocation Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/60—Software-defined switches
- H04L49/602—Multilayer or multiprotocol switching, e.g. IP switching
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/20—Support for services
- H04L49/201—Multicast operation; Broadcast operation
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/25—Routing or path finding in a switch fabric
- H04L49/253—Routing or path finding in a switch fabric using establishment or release of connections between ports
- H04L49/254—Centralised controller, i.e. arbitration or scheduling
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L49/00—Packet switching elements
- H04L49/35—Switches specially adapted for specific applications
- H04L49/351—Switches specially adapted for specific applications for local area network [LAN], e.g. Ethernet switches
Definitions
- Invention relates to digital networks, particularly to multi-layer switching network apparatus and method.
- LAN local area network
- TCP/IP TCP/IP
- LAN size As businesses increasingly rely on such technologies, both LAN size and TCP/IP traffic volume that runs across them have grown dramatically. This has led the network manager on continuous search for products to increase network performance, easily adapt to changing network requirements, and preserve existing network investment.
- LAN technology is evolving into Gigabit per second (Gbps) range.
- Equipment designers have been challenged to make network interfaces and networking products such as bridges, routers, and switches, fast enough to take advantage of the new performance.
- Compounding the equipment design problem has been the rapid innovation in networking protocols.
- the traditional response to this shifting sands problem has been to build easily upgradable software-intensive products.
- these software intensive products typically exhibit poor system performance.
- Invention resides in a multilayer switching device and associated technique for enabling simultaneous wire-speed routing at layer 3 , wire-speed switching at layer 2 , and support multiple interfaces at layer 1 , according to OSI reference model.
- Inventive implementation may be embodied using one or more integrated circuits (ASIC), RISC processor, and software, thereby providing wire-speed performance on interfaces, in various operational modes.
- ASIC integrated circuits
- RISC RISC processor
- FIG. 1 is system-level diagram of preferred embodiment.
- FIGS. 2 A-B are block diagrams of first- and second-level switch respectively of present embodiment.
- FIG. 3 is general switch block diagram of present embodiment.
- FIG. 4 is general control-path diagram of present embodiment.
- FIG. 5 is general datapath diagram of present embodiment.
- FIGS. 6 A-B are block diagrams of LAN interface and datapath interface respectively of present embodiment.
- FIGS. 7 A-B are block diagrams of DMA transfer between local memory and packet memory, and processor access to packet memory respectively of present embodiment.
- FIGS. 8 A-B are processor access to L 3 CAM memory and control memory respectively of present embodiment.
- FIGS. 9 A-B are processor access to L 2 CAM memory, and LAN arbiter interaction with datapath respectively of present embodiment.
- FIGS. 10 A-B are transmit queue management block (XQMB) interfaces and operation respectively of present embodiment.
- XQMB transmit queue management block
- FIG. 11 is DMA block diagram of present embodiment.
- FIG. 12 is flowchart of CPU-to-packet memory operation of present embodiment.
- FIG. 13 is flowchart of packet memory-to-CPU operation of present embodiment.
- FIG. 14 is block diagram of L 3 block interfaces of present embodiment.
- FIGS. 15 A-B are flowcharts of age table maintenance of present embodiment.
- FIGS. 16 A-B are flowcharts of search and lookup operations respectively of present embodiment.
- FIG. 17 is flowchart of packet reception of present embodiment.
- FIG. 1 is top-level overview diagram of system architecture for preferred embodiment.
- Multilayer switch device 6 couples local area network (LAN) workgroup hubs 2 through enterprise switching hub 4 to wide-area network (WAN) links through multiprotocol router 8 .
- Multilayer switch 6 and associated technique enables simultaneous wire-speed routing at Layer 3 (L 3 ), wire-speed switching at Layer 2 (L 2 ), and support multiple interfaces at Layer 1 (L 1 ), according to OSI reference model.
- System may be embodied using one or more integrated circuits (ASIC), RISC processor, and software, thereby providing wire-speed performance on various interfaces in various operational modes.
- ASIC integrated circuits
- RISC processor RISC processor
- System architecture comprises two-level distributed multilayer switch, preferably using 4-Gbps non-blocking switch fabric 6 .
- Multilayer (i.e., both L 2 and L 3 ) switch fabric is entirely contained within single ASIC capable of switching 3M pps or more.
- a 4 Gbps I/O bus connects one or more interface modules to the ASIC.
- the switch matrix is not necessarily integrated with the MAC layer, a wide range of interface types can be supported (i.e., both LAN and WAN).
- various combinations of layer 1 interfaces are supportable, and all interface modules are field-upgradable.
- Various interface modules may carry multiple physical interfaces.
- first-level switch 22 includes switch ASIC 20 , which couples RISC coprocessors (i.e., Network Management Processor (NMP) 10 and Route/Switch (RS) processor 12 ,) for supporting for higher-layer software functions and support features. Optional components may be added for redundancy of critical system components, such as power supplies. Memory 16 and input/output (I/O) modules 14 couple to switch circuit 20 .
- RISC coprocessors i.e., Network Management Processor (NMP) 10 and Route/Switch (RS) processor 12 ,
- NMP Network Management Processor
- RS Route/Switch
- Memory 16 and input/output (I/O) modules 14 couple to switch circuit 20 .
- second-level switch or cross-bar interconnection 18 couples multiple first-level switches 22 .
- aggregate performance of non-blocking switch fabric may exceed 24 Gbps.
- RISC processors 10 , 12 provided in each switch element 22 execute software to provide standards-based dynamic routing, and non-real time activities such as network management.
- Software is stored in flash memory, and is network-updatable via TFTP.
- Preferred software functions include: dynamic Internet Protocol (IP) routing (e.g., RIP, RIPv2, OSPF); layer 2 support (e.g., 802. ID STP); configuration support (e.g., enable/disable Layer 2 or Layer 3 support on per-port basis; ports can be grouped into broadcast domains, flexible subnet configuration); network management: (e.g., SNMP, HTML, Telnet, TFTP, DHCP support).
- IP Internet Protocol
- layer 2 support e.g., 802. ID STP
- configuration support e.g., enable/disable Layer 2 or Layer 3 support on per-port basis; ports can be grouped into broadcast domains, flexible subnet configuration
- network management e.g., SNMP, HTML, Telnet, TF
- Additional software functions include: quality-of-service provisioning (QOS) (e.g., providing multiple levels of prioritization, address- and policy-based QOS, dynamic layer 3 , QOS based on RSVP); IP Multicast (e.g., IGMP, DVMRP); network traffic monitoring (e.g., RMON); hot standby support (e.g., VRRP); additional dynamic routing (e.g., NHRP); and certain IEEE enhancements (e.g., 802.IQ (i.e., VLAN), 802.3x (i.e., flow control), and 802.1p (i.e., priority)).
- QOS quality-of-service provisioning
- IP Multicast e.g., IGMP, DVMRP
- network traffic monitoring e.g., RMON
- hot standby support e.g., VRRP
- additional dynamic routing e.g., NHRP
- certain IEEE enhancements e.g., 802.IQ (i.e., VLAN), 802.3x (i.
- present multilayer switch system is suitable for applications at network aggregation points
- present system may also be used in high-performance workgroup and server applications.
- present system may interconnect between cluster of closely cooperating high performance computers, such as in video postproduction, where ability to transfer data rapidly between workstations is critical to production throughput.
- wire-speed performance is interesting, and flexible layer 3 -addressing support provides connections outside workgroup, without impacting switching speed.
- present multilayer switch system provides network attachment point for one or more servers. Wire-speed performance of present system allows network designer to use either layer 2 or layer 3 topologies, and removes potential network performance bottleneck.
- preferred implementation of innovative multilayer switch apparatus and methodology provides following functionality: support for 16 or more full-duplex 100BaseT ports or up to 28 ports of 10/100BaseT ports; direct interface to MIPS-type RISC processor for management and routing; integration of SDRAM controller for shared high-speed 6-channel packet memory; integrates of CAM access interface to system processor; integration of hardware CAM processor for L 2 learning, lookup and live interactive activities or transactions; integration of hardware hash-based IP header lookup and management; integration of hardware-based transmit and free queue management; integration of L 2 and L 3 forwarding of unicast, broadcast and multicast packets; broadcast traffic management; integration of QoS, with 4 priority queues per port; hardware-handled packet movement; integration of 768 bytes of dual-port memory for L 2 and L 3 header for 28 ports; support for 4MB/16MB of SDRAM packet memory; implementation of 256 bytes of data buffers for concurrent transfers to PM SDRAM and LAN bus; intelligent buffer scheduler & arbiter
- multilayer switch circuit 20 is implemented as single-chip integrated circuit (e.g., semicustom ASIC) for processing handles switching of any canonical packet, cell, frame, or other data communication element, with no or limited processing assistance from external processors.
- Switch circuit 20 operates in relatively low latency, and store-and-forward switching modes. Transactions between Ethernet ports may operate in low-latency cut-thru mode; other transactions may occur in store-and-forward mode.
- switch circuit 20 may contain substantially one or more of following functions: external bus interface, processor interface, CAM interface, LAN interface, packet memory (PM) SDRAM interface, route cache SDRAM interface, control memory (CM) SRAM interface, LAN block, LAN bus arbiter, LAN bus controller, LAN block interfaces, data path block, data path buffers, data path controller, buffer scheduler, packet memory, packet memory SDRAM arbiter and controller, DMA function-to-processor interface, packet engine (PE), port control function, port attribute memory, L 2 CAM engine, memory blocks for header and CAM analysis result, CAM structures, L 2 header analysis hardware engine, auto-forwarding block, forwarding block, L 3 header analysis result memory, free queue management block, block attributes management, transmit queue management block (XQMB), SRAM arbiter and controller, processor interface, L 3 block, L 3 header memory, hash function, L 3 lookup algorithm, L 3 management function, L 3 aging function, route cache (RC) SDRAM arbiter and controller, RISC processor interface
- switch system shows general logic block diagram for switch circuit 20 coupled to: 64-bit 66 Mhz LAN bus, external memory 16 through 32-bit 99-Mhz bus, L 2 CAM through 16-bit 66-Mhz bus, control memory 136 through 16-bit 66-Mhz bus, L 3 route cache through 16-bit 66-Mhz bus, and switch processor 12 through 16-bit 66-Mhz bus, which couples to network management processor (NMW) 10 through external interprocessor controller (IPC) 24 .
- NMW network management processor
- IPC interprocessor controller
- internal control path of switch circuit 20 is shown.
- External switch processor 12 couples to CAM interface 46 , free queue management 48 , L 3 lookup 50 , transmit queue management and scheduler 58 , SDRAM memory controller 62 , and SRAM memory controller 64 .
- internal control path includes forwarding engine 52 , which couples to CAM interface 46 , free queue management 48 , L 3 lookup 50 , block attributes 60 , transmit queue management and scheduler 58 , and receive block 54 .
- Transmit queue management and scheduler 58 couples to transmit block 56 , SRAM memory controller 64 , and block attributes 60 .
- Receive block 54 and transmit block 56 couple to LAN bus.
- CAM interface 46 couple to CAM bus and receive block 54 .
- SRAM memory controller 64 couples to free queue management 48 , block attributes 60 , L 3 lookup 50 , and SDRAM memory controller 62 .
- SDRAM memory controller 62 couples to RC memory bus and L 3 lookup 50 .
- Block attributes 60 couples to free queue management 48 .
- Forwarding engine 52 couples to receive block 54 .
- multi-channel packet memory arbiter and controller 66 couples to SDRAM packet memory bus, processor and DMA interface 68 , L 3 engine 70 , receiver buffers 72 , and transmit buffers 74 .
- Receive and transmit buffers 72 , 74 couple to media access controller (MAC) first-in first-out (FIFO) bus.
- MAC media access controller
- FIFO media access controller
- Switch circuit 20 includes processor interface 36 which couples to 32-bit MIPS RISC processor multiplexed bus (e.g., NEC R4300).
- processor bus a 32-bit address/data bus operable up to 66 Mhz, operates in master and slave modes.
- slave mode such processor bus responds to accesses to internal resources, such as registers, CAM 142 , Control Memory 136 , PM SDRAM and RC SDRAM.
- master mode such bus handles DMA operations to and from PM SDRAM.
- Such processor bus does not respond to accesses to external resources, but cooperates with external system controller circuit.
- master mode such processor bus may handle DMA to system memory.
- Switch circuit 20 includes CAM interface 46 , a dedicated 16-bit bus compliant with content-addressable memory (i.e., Music Semiconductor CAM 1480 compatible) operating at 66 Mhz. Such bus may be shared by external interface.
- content-addressable memory i.e., Music Semiconductor CAM 1480 compatible
- Such bus may be shared by external interface.
- RS route/switch
- Switch circuit 20 generates CAM access timing control on behalf of RS processor 12 .
- Switch circuit 20 learns and looks-up MAC addresses and port numbers through such bus.
- Switch circuit 20 includes LAN interface 40 which couples LAN bus, a 64-bit access bus operating at 66 Mhz. Ethernet MAC devices connect to such LAN bus through receive and transmit MAC FIFO bus. Switch circuit 20 generates select signals and control signals for access to external MAC device FIFO bus. Switch circuit 20 reads writes data in 64-bit single-cycle burst mode. Burst size is 64 bytes. Preferred bandwidth is 4 GB/s at 64-bit/66 Mhz-operation at 64-byte slice size. Ethernet frames are transferred across LAN bus. At end of receive frame, status bytes are read.
- Switch circuit 20 includes packet memory (PM) SDRAM interface 42 , which includes PM SDRAM bus which operates at 32-bit/99-Mhz standard. Packet memory 16 is directly connected to such bus through registered transceivers. Preferred bandwidth is 400 MB/s at 99-Mhz operation and 64-byte burst mode. Seven-channel arbiter inside switch circuit 20 allows up to 7 agents to access packet memory 16 . PM interface supports up to 8 MB of SDRAM in two banks.
- PM SDRAM interface 42 includes PM SDRAM bus which operates at 32-bit/99-Mhz standard. Packet memory 16 is directly connected to such bus through registered transceivers. Preferred bandwidth is 400 MB/s at 99-Mhz operation and 64-byte burst mode. Seven-channel arbiter inside switch circuit 20 allows up to 7 agents to access packet memory 16 . PM interface supports up to 8 MB of SDRAM in two banks.
- Switch circuit 20 includes interface to Route Cache (RC) SDRAM for coupling timing control signals and multiplexed 16-bit bus, which operates in 66-Mhz mode capable of streaming data at 132 MB/sec.
- RC Route Cache
- Switch circuit 20 includes interface to Control Memory (CM) SRAM for managing block free queue list, transmit queues, block parameters and L 3 CAM aging information.
- CM Control Memory
- Such interface is 16-bits wide and operates at 66-Mhz. Address and data buses are multiplexed and operate in flow-through and pipelined modes.
- FIG. 6A shows LAN block and interfaces 40 externally to Ethernet Media Access Controller (MAC) FIFO bus and internally to CAM interface block, datapath block 44 , and packet engine block 82 .
- LAN block interface functionality include bus arbitration for receive and transmit requests of FIFO bus, bus control and protocol handling, signaling internal datapath block to initiate data transfers and communicating with packet engine to signal begin and end of receive and transmit operations on FIFO bus.
- datapath block 44 couples to FIFO data bus, LAN bus controller 76 , buffer allocator 78 , and packet engine 82 .
- LAN bus controller (LBC) 76 couples to FIFO bus control, buffer allocator 78 , and receiver and transmit arbiters 80 , which couple to packet engine 82 and receive and transmit requests.
- LBC LAN bus controller
- LAN interface 40 When LAN interface 40 operates, receive requests and transmit requests are multiplexed and fed by external logic. Multiplexer uses 2-bit counter output. Front end demultiplexer reconstructs requests on 32-bit receive request register and 32-bit transmit request register. Few clocks latency for request may be sensed to be activated or deactivated, which may be handled by arbiter mechanism 80 .
- Receive arbiter 80 services receive port requests, preferably in round-robin scheme for equal distribution. Overlapped processing provides improved performance. Hence, if receive port is under service, next request prioritization occurs in parallel. During arbitration, arbiter 80 may receive port enabled, free block allocated signals from other modules. Upon certain channel winning arbitration, internal receive buffer is allocated 78 , and data staged from MAC FIFO bus for packet memory 16 . When buffer is granted, channel is presented to LAN Bus controller 76 for data transfer.
- transmit arbiter 80 services transmit port requests in round-robin scheme for equal distribution. Overlapped processing provides improved performance. Hence, when transmit port is under service, next request is prioritized in pipeline.
- arbiter 80 may receive port enabled, valid packet assigned, in link mode the transmitter has at least one slice signals from other modules. If channel has data slice in datapath 44 , channel is not allowed to join arbitration until data is put into packet memory 16 , thereby preventing out-of-sequence data transfer.
- Upon channel winning arbitration it is presented to buffer allocator block 78 to obtain internal transmit buffers for staging from packet memory 16 for MAC FIFO bus. Once transmit request wins arbitration, and transmit buffer is allocated, channel is presented to packet engine block 82 to obtain data from packet memory 16 . Once data is staged in transmit buffer, buffer requests to LAN Bus controller 76 to transfer data in transmit buffer to MAC FIFO bus.
- LAN bus controller 76 provides access to MAC FIFO bus targeted to port moving slice between MAC FIFO and internal data buffers. Receive request, which wins receive arbitration and secures one of receive buffers from buffer allocator 78 and transmit buffers having data for transfer to FIFO bus, competes for services of LAN bus controller 76 . Arbitration mechanism is configured to split bandwidth evenly between receive requests and transmit requests. LAN bus controller 76 generates end-of-packet status read cycles for receive request data transfer operations. Status information is used to determine if received packet is good or bad. If error is sensed, received packet may be rejected.
- Data bus width of LAN bus is 64 bits.
- LAN bus access is performed in burst mode (i.e., single-cycle burst mode) with maximum of 64-byte transfer, preferably executing at 8 data cycles in burst.
- LAN bus controller 76 is started by buffer scheduler when data buffer is allocated to receive or when data transfer from packet memory 16 to one of transmit buffers is complete.
- Receive and transmit data to LAN bus is staged through 64-byte deep receive and transmit data buffers in datapath block 44 .
- Receive and transmit requests arbitration and FIFO bus control are handled by LAN block.
- Buffer allocator 78 in datapath block 44 manages allocation of receive and transmit buffers, and packet engine block 82 handles movement of data between packet memory 16 and receive and transmit buffers.
- FIG. 6B shows datapath block 44 interface, including packet memory controller 82 coupled to data buffers 84 and packet memory engine (PME) 90 .
- Data buffers 84 couple to LAN block 86 , buffer scheduler 94 , slice counters 88 .
- Buffer attributes 92 couple to PME 90 and LAN block 86 , which couple to buffer scheduler 94 .
- Data transfers between packet memory bus and MAC FIFO bus are staged through receive and transmit buffers 84 in datapath block 44 .
- Block logic tracks state of buffers 84 .
- Datapath block 44 interacts with LAN Block 86 , packet engine block 82 and packet memory controller 82 .
- Two buffers are provided for transmission, and two buffers are provided for reception. Such buffers are associated with respective buffer status. Transmit buffers hold data form PM 16 to MAC FIFO (LAN) bus. Receive buffers hold data from MAC FIFO bus to PM 16 . Each buffer has dedicated channel to PM SDRAM Controller. PM SDRAM Controller arbitrates each request to transfer on first-come/first-serve basis. On LAN side, appropriate buffer is selected for read or write.
- Frame transfer across LAN bus occurs on slice basis.
- Slice is 64 bytes.
- burst data transfer size is slice size, except for last slice in frame.
- Last slice size is decided by frame size. Ports are serviced, in time-division multiplex mode.
- Receive slice buffer is used to capture LAN data from MAC FIFO.
- Slice is 64 bytes.
- Switch circuit 20 has two 64-byte buffers.
- incoming 64-bit data words are strobed on selected slice buffers, word-by-word, during clock edges. Write order is from top to down.
- Receive status is maintained for respective receive slice. For example, slice status provides:
- Receive slice size (represented by 6-bit number.) Maximum is 64 bytes.
- MAC provides in each data phase, valid bytes through bits (e.g., LBE# ⁇ 7-0>). Hence, LBEI# ⁇ 7-0> are registered and analyzed at end of data phase to provide cumulative slice size.
- MAC provides in each read data phase, if end-of-frame.
- EOFI# signal is registered and stored for EOF status. It is also used to close current transfer.
- SOF signaling The MAC provides on each read data phase, if Start-of-frame. SOFI# signal is registered and stored for SOF status.
- Transmit slice buffer is used to capture (e.g., PMDO) bus data and supply to LAN bus.
- Slice is 64-bytes.
- Switch circuit has two 64-byte slice buffers.
- 64-bit data words are read from selected slice buffer.
- One clock pre-read is implemented to provide minimum delay time on LAN data (LD) bus. Read order is from top to down.
- LD LAN data
- Status is maintained for respective transmit slice.
- Slice status is loaded by PM engine 90 when moving slice from PM.
- Status information includes:
- Slice size (represented by 6-bit number.) Maximum is 64 bytes.
- PM engine registers slice size.
- PM engine 90 registers signal while transferring slice from PM bus. If status is on, LAN FIFO controller asserts EOF# signal at appropriate data phase.
- Buffer scheduler 94 allocates transmit and receive data buffers to requesting agents, keeps track of busy/free status of each buffer, and allocates free buffer to requesting agent. Buffer scheduler 94 optimizes for (a) equal distribution of bandwidth between receivers and transmitters, (b) avoiding deadlock situation of transmit buffer, and (c) achieving highest concurrence of LAN bus and PM bus.
- Datapath controller includes buffer attributes 92 for receive and transmit buffers 84 , and track byte count per slice basis.
- Buffer attributes 92 such as End-of-Packet (EOF), start-of-packet (SOF), Byte Enables (BEB), and Slice Count are tracked from time data arrives into receive or transmit buffer until data leaves buffer.
- Buffer attribute 92 information is used by packet memory engine 90 to track progress of packet flowing through switch circuit 20 per slice basis.
- Datapath controller interacts with buffer scheduler 94 at end of slice transfer to release buffer. Synchronization between PM SDRAM controller and LAN bus interface 40 is thereby accomplished.
- Packet memory resides on dedicated SDRAM bus.
- Switch circuit 20 integrates SDRAM controller to access packet memory 16 .
- PM SDRAM controller functionality includes: 32-bit interface operating at 99-Mhz to 8 MB of external SDRAM; support for up to 7 internal requesting agents; arbitrates requests and generates request to SDRAM control block; pipelines requests for maximum efficiency and throughput; bursts of 4 (one bank), 8 or 16 (both banks) accesses on SDRAM; and maximum performance at 16 bursts and minimum performance at single read or write.
- Route processing is provided by MIPS R4000 family RISC processor 12 , which interfaces with switch circuit through address/data multiplexed bus.
- RISC processor interface may use external system controller, for example, for communicating with switch circuit 20 though processor slave port.
- RISC processor serves switch or route processor 12 .
- Several register resources in switch circuit 20 are used by RISC processor 12 to control configuration and operation of switch circuit 20 .
- RISC processor 20 may access resources outside of switch circuit 20 , such access being controlled by switch circuit 20 packet memory 16 , route cache memory, and CAM for L 2 forwarding.
- Switch circuit 20 communicates status of operation and draws attention of processor 12 through status and process attention registers. When configured, switch circuit 20 performs DMA of data from packet memory to processor local memory, and forwards packets to processor queue maintained by switch circuit 20 .
- route processor (RP) 12 is NEC Vr4300 RISC microprocessor from MIPS family with internal operating frequency of 133 Mhz and system bus frequency of 66 Mhz.
- Processor 12 has 32-bit address/data multiplexed bus, 5-bit command bus for processor requests and data identification, six handshake signals for communication with external agents, and five interrupts. Bus width can be selected as 32-bit operation.
- Processor 12 supports 1, 2, 3 and 4-byte single accesses and 2, 4 and 8 word burst accesses. Processor 12 uses little endian when accessing switch resources.
- RP 12 is interfaced to switch circuit 20 .
- RP 12 communicates with NMP 10 through interprocessor communication (IPC) bus 24 , and accesses switch local resources, such as packet memory 16 , L 3 CAM (Route Cache) 28 , control memory 136 and L 2 CAM 142 through switch circuit 20 and local resources, such as local memory, ROM etc., through system controller.
- IPC interprocessor communication
- switch local resources such as packet memory 16 , L 3 CAM (Route Cache) 28 , control memory 136 and L 2 CAM 142
- switch circuit 20 and local resources, such as local memory, ROM etc.
- Two interrupts are used by switch circuit 20 to issue interrupt requests to processor 12 .
- Two slaves on RP processor 12 system bus are switch and system controller.
- Switch is final agent to provide ready signal to processor requests that switch or system controller is ready to accept. During DMA transfer, switch acts as master.
- Write access is implemented as ‘dump and run’ with two pipelined buffers to improve system performance. This allows two back-to-back write cycles.
- One read request is processed at a time.
- Processor 12 accesses internal registers resources in 32-bit mode.
- Write buffer and read buffer are provided to packet memory 16 to match frequency difference of 99-Mhz and 66-Mhz.
- Memory interface to switch is 32-bit.
- Maximum burst size to packet memory 16 is four 32-bit words (i.e., 16 bytes).
- Read buffers are provided to L 3 CAM and control memory 136 because of 16-bit interface to switch. Little endian is used when data is packed and unpacked during write or read requests to 16-bit interfaced memories.
- Maximum burst size to L 3 CAM 28 is 16 bytes, and to CM is 8 bytes.
- Write or read request to memories is arbitrated through agents inside switch, such as forwarding engine, L 3 engine etc., so latency depends on various factors.
- processor 12 owns mastership or control of bus.
- processor 12 enters into uncompelled slave state after address phase, giving bus control to external agent to drive data.
- FIG. 7A illustrates DMA transfer between RP processor 12 local memory 100 and packet memory 16 .
- DMA transfer between packet memory 16 and NMP processor local memory is also provided in architecture.
- NMP processor system controller responds to DMA master requests between packet memory and NMP processor local memory.
- DMA is implemented using two design blocks called DMA engine 104 and DMA master 102 .
- DMA engine is interfaced to packet memory 16 and that of DMA Master to processor system bus.
- DMA is initiated by setting bits in DMA command register.
- DMA transfer between local memory 100 and packet memory 16 occurs substantially as follows:
- DMA engine 104 notifies DMA master 102 to initiate DMA transfer when packet is pending by giving request.
- DMA master 102 arbitrates for processor bus with RP processor 12 as another master by giving request (e.g., EREQ) to processor 12 .
- request e.g., EREQ
- switch circuit 20 acts as master to system controller 98 .
- Processor gives bus control to RP processor when ready.
- bus is granted by processor, DMA transfer begins. Mastership of processor bus can be re-acquired by RP processor 12 between each slice transfer, which is maximum of eight 32-bit words (i.e, 32 bytes).
- DMA engine 104 reasserts request after each slice transfer, until block of packet data is transferred. At end of DMA, bus control is given to processor.
- DMA master 102 When bus is in uncompelled slave state, DMA master 102 does not access processor system bus to simplify design. While DMA transfer is taking place on bus, system controller 98 does not drive bus, assuming bus in slave state.
- FIG. 7B illustrates RP processor 12 access to packet memory (PM) 16 through L 2 /L 3 switch circuit 20 .
- Switch interface to packet memory 16 is 32-bit, and maximum burst size is 16 bytes.
- Synchronous DRAM is chosen for packet memory that can be operated at 66-Mhz, 99-Mhz and 125-Mhz frequencies.
- processor write request processor dumps write-data into front-end pipeline buffers 106 .
- Slave state machine 108 provides such data into packet memory write buffer 110 .
- Processor request is arbitrated with LAN requests and L 3 engine requests in PM SDRAM arbiter to access PM 16 .
- PM SDRAM controller 112 generates control signals for SDRAM.
- PM read buffer 114 During processor read request, read-data is provided in PM read buffer 114 from packet memory bus. Synchronizer 116 converts 99-Mhz signal into 66-Mhz pulse signal that initiates slave state machine to empty read buffer. Read data is muxed with data from other blocks and driven to processor system bus. Packet memory to local memory (PM-to-LM) DMA transfer data is not written into read buffer, but passed to processor system bus.
- PM-to-LM Packet memory to local memory
- FIG. 8A illustrates RP processor 12 access to L 3 CAM (route cache) memory 28 .
- RP processor 12 accesses L 3 CAM 28 through switch circuit 20 to initialize entries and maintain data structures.
- FIG. 8B shows control memory 136 access through switch circuit 20 .
- RP processor 12 couples to switch circuit 20 through 66-Mhz, 32-bit processor system bus, wherein pipeline buffers 106 receive processor write data and couple to slave state machine 108 .
- L 3 CAM write buffer couples to slave machine 108 and L 3 CAM SDRAM controller 120 , which receive requests from other agents and couples to L 3 CAM memory 28 through 66-Mhz, 16-bit bus.
- L 3 CAM read buffer 122 provides read data through 32-bit processor bus and couples to slave state machine 108 and register 134 over 16-bit bus.
- Register 134 receives 66-Mhz clock signal and couples to L 3 CAM memory 28 through 66-Mhz, 16-bit bus.
- CM write buffer 128 couples to slave machine 108 and CM SSRAM controller 130 , which receive requests from other agents and couples to control memory 136 through 66-Mhz, 16-bit bus.
- CM read buffer 132 provides read data through 32-bit processor bus and couples to slave state machine 108 and register 124 over 16-bit bus.
- Register 124 receives 66-Mhz clock signal and couples to control memory 136 through 66-Mhz, 16-bit bus.
- Synchronous SDRAM is chosen for L 3 CAM 28
- Synchronous SRAM is chosen for control memory 136 .
- Switch interface to both memories is 16-bit, and both memories operate at 66-Mhz.
- Processor 12 access to memories is similar in both cases, maximum burst size to L 3 CAM memory is 16 bytes, and maximum burst size for control memory 136 is 8 bytes. Data is packed and unpacked for each processor access.
- Each memory 28 , 136 has write buffer 118 into which processor write-data is provided from pipeline buffers 106 by slave state machine 108 . Since memory interface is 16-bit, processor write data is divided into two 16-bit half words.
- Processor 12 request to L 3 CAM memory 28 is arbitrated with L 3 engine requests.
- Processor 12 request to control memory is arbitrated with forwarding engine, FQMB, L 3 engine, XQMB, BAM and DMA.
- FQMB forwarding engine
- L 3 engine L 3 engine
- XQMB XQMB
- BAM DMA
- DMA digital multiple access memory
- L 3 CAM or CM read data from memory bus is provided in read buffer.
- slave state machine 108 starts emptying read buffer 122 appropriately and packs two half words into 32-bit word, and puts on processor system bus.
- RP processor 12 accesses L 2 CAM memory 142 through switch circuit 20 .
- Content Addressable Memory CAM
- Switch circuit 20 interface to L 2 CAM memory is 16-bit.
- Processor 12 executes commands write/read and data write/read to L 2 CAM 142 using CAM access registers provided inside switch circuit 20 .
- Processor 12 accesses L 2 CAM 142 through register-based request-grant handshake by loading L 2 CAM Access Control & Status Register to execute processing cycles.
- RP processor 12 arbitrates with CAM arbiter 138 in switch circuit 20 for CAM bus.
- slave state machine 108 For processor requests, slave state machine 108 generates control signals for CAM 142 , and CAM arbiter engine 138 processes switch requests.
- processor 12 provides write-data in pipeline buffers 106 .
- slave state machine 108 puts data from pipeline buffer 106 on CAM bus.
- read-data from CAM bus is muxed 140 with data from other blocks and passed to processor system bus. Write/read buffers need not be provided in present case.
- FIG. 9B shows LAN arbiters interaction with datapath.
- Register files 144 for receivers and transmitters including corresponding block address registers and byte count registers, couple to block address register and byte count register coupled to state machines 148 , which couple to switch data path 44 and receive and transmit arbiter 150 .
- Packet switch engine 82 performs control functions for transfer request in and out of receive and transmit buffer to packet memory 16 . Packet engine 82 handles LAN data movement, command process, and PM address calculation.
- packet memory engine 82 sets up for moving slice between packet memory 16 and allocated data buffer. This is triggered by scheduler while slice is scheduled to move in/out of data buffer. PM engine has access right to block address registers and registers to understand actual address on PM 16 and update packet size.
- packet memory engine 82 executes systematic hardware processes when Forwarding Block and Transmit Queue Management Block (FB/TQMB) generates instructions such as: link, receive enable, transmit enable, receive reject, etc. Hence, end of packet reception/transmission is noticed for next packet initialization. In notifying such events, priority encoding is provided for first-in/first-service integrity.
- FB/TQMB Forwarding Block and Transmit Queue Management Block
- packet memory engine 82 regards Ethernet ports as 32 concurrent full-duplex DMA channels. Relevant PM pointers for each channel are maintained. For every slice transfer, PM bus address is calculated.
- buffer attributes 92 are provided in attribute block address array, which is 3-port architecture having 64 ⁇ 12-bit 3-port memory array.
- Port-i is write port; port-2 is read port; and port-3 is read port.
- Packet memory engine 82 can write/read memory locations using various ports.
- Forwarding engine (FE) can read locations asynchronously.
- Port-3 is assigned for FE.
- First 32 locations are used for “Receive block address” of 32 receive ports.
- Next 32 locations are “Transmit block address” for 32 transmit ports.
- PM Engine 82 initializes block address for receive/transmit ports on command of Auto Forwarding Block.
- PM 10 engine 82 reads block address relevant to receive/transmit port under service.
- PM engine 82 uses block address to identify packet in PM 16 .
- CAM interface block analyzes incoming packet at layer 2 , i.e., at MAC layer. Analysis result is forwarded to Auto Forwarding Block state machine. CAM processor is called for attention when ether header block is loaded in ether header memory. On such trigger condition, after acquiring CAM bus interface, CAM Processor starts defined fast processing action. Block contains layer 2 header memory, analyzed and to be analyzed port FIFOs, and result memory. CAM block interfaces to internal memories organized as memories and FIFOs as well as external CAM to accomplish L 2 lookup.
- begin of header is identified and required header information is loaded into ether header memory. Sixteen-byte header blocks are reserved for each port in header memory. Loaded indication is updated on 5-bit entry in 32-deep ether header to-be-analyzed FIFO. Such FIFO provides first-in/first-service feature.
- Ether header memory is 2-port memory has 64 ⁇ 64 bit architecture. Port- I is write port, and port-2 is read port. Such memory is located on LAN side of receive buffer. As first slice of new receive packet is loaded into receive buffer, header slice (i.e., 16 bytes) is written to ether header memory in 64-bit words. Ether port number is used as reference address to select header block number. Maximum of 32 header blocks can be stored in such memory. Port-2 is used by CAM processing engine. CAM engine reads 16-bit quantity at a time through front-end 64:16 multiplexer. L 3 header information, up to 8 bytes per port is stored in different memory. Such information is used by L 3 lookup engine during routing operation.
- Ether-to-be-analyzed FIFO memory is 32 ⁇ 5-bit two-port memory, holding maximum of 32 port numbers to be analyzed.
- Port-1 is write port
- port-2 is read port.
- FIFO is written with port number when first slice of data is received on LAN bus and header loaded in ether header memory.
- CAM Processor reads port number through port-2 for indexing header memory.
- FIFO structure ensures that ports to be analyzed are presented to CAM engine in arrived order.
- Ether analyzed FIFO memory is 32 ⁇ 6-bit two-port memory, holding maximum 32 analyzed port numbers.
- Port-1 is write port
- port-2 is read port.
- CAM Processor writes analyzed port number through port-1 and Forwarding Engine (FE) reads through port-2.
- FIFO structure ensures that analyzed ports are presented to forwarding engine in arrived order.
- Ether result memory is 32 ⁇ 16-bit two-port memory, holding results for 32 ether ports.
- Port-1 is write port
- port-2 is read port.
- CAM Processor writes L 2 forwarding result through port-1
- FB Forwarding Block
- FB reads port number from Analyzed FIFO to make forwarding decision.
- FB uses port number as reference address to read CAM analysis result.
- External CAM memory is 1024 ⁇ 64-bit capacity on standard configuration. Size can be expanded to 2048 ⁇ 64-bit by adding CAM device in vertical expansion.
- CAM memory is connected on dedicated CAM bus. Such bus is shared between CPU and switch circuit. Normally such bus is default-owned by switch circuit. CPU can use bus by register mode bus handshake.
- CAM memory contains 1024 locations of 64 bits wide. Locations can be assigned as RAM property or CAM property. Location assigned as RAM will not be accounted in lookup process; CAM locations participate in lookup process. Repeatedly used parameters are stored in RAM so that real-time data movement between RAM location and Comparand/mask registers/etc. can happen with minimum overhead. Every location has associated status filed which describes entry, such as: empty entry, valid entry, skip entry on lookup, and RAM entry.
- Layer 2 header analysis is performed by CAM processor. Ethernet headers are loaded and processed on dedicated Ethernet header memory having 128 ⁇ 32 bit dual port memory. Assuming case where packet received on port(x), switch circuit 20 is triggered on such packet by request from MAC port number(x), which is effectively hardware packet arrival notification.
- Header is extracted from MAC received data stream. New receive packet data is identified with arrival of SOF, and first 16 bytes are treated as layer- 2 header. If header is concurrent to store access to receive buffer, then header is stored in port-specific block number(x) in header memory. Writing process may not account for port contention. Block written on port-1 may not be accessed on port-2. Header is stored as header block(x). At end of storage, port number is written in ether-to-be-analyzed FIFO, which is 32 ⁇ 5-bit register. FIFO write pointer is incremented after each status write
- CAM processor starts when valid entry is loaded in Ether-to-be-analyzed FIFO. CAM Processor maintains read pointer to read valid entry. Valid entry is notified if there is difference between write pointer and read pointer. Entry read provides port number of header. CAM Processor uses port number to reach header block(x).
- switch system has 32 ports and 32 entries. New packet on port can not be received unless old packet is processed, according to system-level handshake. Hence, at any time, no more than 32 headers/header status may be stored, effectively reducing complexity of FIFO design. PM engine can blindly write header/status without looking for FIFO-full condition. CAM Processor can start as long as pointers are not equal.
- CAM processor handles header processing. CAM processor is notified of Ethernet header valid when write pointer and read pointer differ. When entry is valid on Ethernet-to-be-analyzed FIFO, CAM processor reads entry and increments read pointer. Using such value, CAM processor can reach specified header block. Ether header memory is divided into 32 blocks. Port number directly provides starting address of header block. Entries in block are consecutive 16 bytes.
- CAM processor processes header block, and writes result on port specific location on Ether result memory. CAM process completion is notified to Auto Forwarding Block through Ethernet result FIFO, which is 32 deep register construction. Each entry is 6-bit wide. Entry is result of CAM memory lookup. If set, destination MAC address indicates CAM hit. Routing tag in header block is valid. If clear, CAM lookup fails; routing tag does not contain valid information.
- CAM processor To write on result FIFO, CAM processor has write pointer, which is 5-bit counter. CAM processor write entry, whereas AFB read entry. When CAM completes process, it writes result entry, and increments write pointer. At last, CAM processor increments Ether header status FIFO read pointer to point to next entry.
- CAM processor header processing includes learning process of: source lookup, source port read, and source learning.
- CAM processor learns MAC addresses arriving from Ethernet ports.
- As associated process of CAM lookup CAM processor determines whether source address was learned previously, i.e., by reading source address from Ether header memory, and writes CAM for lookup. If match occurs, processor presumes source port was learned; it reads existing port information from associated data to compare whether port is same as receiving port. If MAC header matches, whether or not ports match, processor makes entry live and at same time relearns receiving port. If receiving port number does not match learned port, Source Address (SA) Learned flag is set. If miss, processor learns entry into next free address if CAM is not full, and if learned SA Learned flag is set. While updating such new entry, processor follows correct data structure for RAM associated information.
- SA Source Address
- attribute is set with (e.g., ETHR_LRN_INHIBIT) register for each port to inhibit learning on specified ports. If set, during source lookup process, after source port read, entry is made live if hit and if miss, MAC address is not learned. Source port read phase can be skipped if source port filtering is not required.
- ETHR_LRN_INHIBIT e.g., ETHR_LRN_INHIBIT
- Destination lookup process includes steps: destination lookup and destination port read.
- CAM processor reads 6-bytes destination MAC address from header memory and writes on CAM for comparison lookup. If miss, destination is assumed unknown; if hit, destination is available through memory (e.g., ARAM) field, which provides destination port number and port/MAC address specific status and control flags. If hit, CAM processor reads ARAM field and writes in result memory, setting hit flag. If miss, CAM processor has nothing to read and write miss flag to result memory. Rest of result data is not valid in miss case. Forwarding blocks read this field for analysis and forwarding decision. At end of process, CAM analysis done flag is set for packet on receiving port.
- CAM processor analyzes results of source lookup and destination lookup processes to determine how to process incoming packet.
- Each port has two bits allocated to handle spanning-tree protocol requirements. One bit is allocated for ‘Port Blocked State’ flag and other for ‘Learn Inhibit’ flag, which is used for learning of MAC addresses on receiving port.
- Port Blocked State’ flag is used for forwarding decision. Filtering bits in result from both source lookup and destination lookup along with port specific STP control bits relating to forwarding, source port and destination port read as result of destination lookup and read are considered.
- CAM processor sets CAM analysis completion status for receiving port. If destination lookup resulted in hit and destination port is one of physical LAN ports and cut-thru switching on port is enabled or CPU port, port number is written to Ether analyzed FIFO. CPU port is allowed to enable Layer 3 analysis parallel to packet reception. Result processing is done by Auto Forwarding Block (AFB). AFB is notified of CAM process completion through Ethernet Analyzed FIFO. AFB can read highest priority FIFO entry using hardware hidden read pointer. If read pointer and write pointer are different, one or more valid entries are available in Ethernet analyzed FIFO. AFB reads valid entry and gets port number. Reading entry increments read pointer, if present entry is valid.
- AFB Auto Forwarding Block
- AFB can access Ether result memory. Refer to Auto Forwarding Block section for details on AFB functionality. If CAM analysis resulted in miss or hit but packet can not be switched, AFB does not need to be notified until packet reception is complete. Hence, CAM processor merely sets CAM analysis completion flag for receiving port. AFB processes packet when both receive completion and CAM analysis completion set for receiving packet.
- Time stamp register provides variable granularity for aging.
- Processor uses instruction set provided by CAM device. Entries to be aged are processed in one instruction, though setup is required before executing instruction. In addition to status bits provided by CAM for every entry, 3 bits in RAM field are dedicated for aging information. Status provide by CAM are used to identify if entry is ‘Valid’, ‘Empty’, ‘Skip’ or ‘RAM only’. One of bits allocated in ARAM field is used to mark entry ‘Permanent’. Entries marked ‘Valid’ and not ‘Permanent’ are considered for aging. Additional two bits in ARAM filed provide flexibility to CPU to implement aging process.
- processor visits CAM to ageout entries, it searches CAM for entries with oldest time stamp. In search process, processor configures mask registers in CAM in such way that age bits enter comparison, and entries that are not ‘Valid’ and marked ‘Permanent’ do not enter comparison. In next instruction, processor can clear ‘Valid’ bits on matching locations to ‘Empty’ state. By doing so, oldest entries are marked empty. From that point, aged entries do not enter compare operation until made ‘Valid’ aging during normal learning process.
- Auto forwarding block is hardware Ethernet packet forwarding engine and queue processor.
- AFB analyzes incoming packet and may forward packet both at layer 2 and layer 3 . After forwarding analysis is done, AFB posts and maintains port queues.
- AFB may accept packets from processor interface and post packet in requested queues.
- AFB provides processing power on packet-by-packet and manages required information for integrity of packet routing strategy.
- AFB feeds initial setup information for each ether packet for each port to run data transaction.
- AFB functionality enables switch circuit 20 to perform forwarding and filtering without real-time assistance from processor 12 .
- Processing element is out of the datapath, and forwarding and filtering is done at line rate for supported ports.
- AFB functionality includes: free queue management, block attributes management, receive port management, forwarding and filtering, transmit queue management, quality-of-service (QoS) support, and control memory interface.
- QoS quality-of-service
- Forwarding function features port linking, wherein receive port is linked to transmit port before packet is fully received, thereby improving latency between received and transmitted packets.
- Port linking is accomplished in forwarding stage if conditions are suitable. For example, packet can cut-thru with unique destination, i.e. no more than one port is target destination for packet. Cut-thru enable bit are satisfied for certain values, such as: destination port, speed-matching logic, xmtPortEn, xmtPortNotBsy, xmtQNotVld and mirrNotVld. Data arrive speed should not exceed transmitting port speed. Transmitter should be ready to accept command.
- Transmitter may be busy transmitting data, or there may be packets waiting for queue transmit Also, there should be minimum of data present in buffer before process can start or arbitration latency may result in transmit FIFO under-run condition. In such case, transmitter is linked but does not start transmitting data until required minimum data is received in packet memory.
- FIG. 10A shows Transmit Queue Management Block (XQMB) 154 , which is hardware block for managing transmit queue functions for switch circuit 20 .
- XQMB 154 couples to forwarding engine (FE) 52 , DMA Engine 104 , block attribute memory (BAM) 152 , PME 90 , queue attribute memory (e.g., AttrRAM) 156 , port AttrRAM 158 , and control memory 136 through interface 160 .
- FE forwarding engine
- BAM block attribute memory
- PME 90 queue attribute memory
- AttrRAM AttrRAM
- port AttrRAM 158 port AttrRAM 158
- control memory 136 control memory
- XQMB 154 functionality includes: initializing and managing transmit queues for each port; maintaining QOS parameters (i.e., tokens) for each port; queues (e.g., nQueues and dQueues) blocks to/from control memory transmitter queues; forwarding blocks to requesting transmitter; returning block numbers to BAM controller 152 ; forwarding multi/broadcast block in ‘background’; supporting 28 physical ports, 3 logical ports and multi/broadcast port; and using round-robin priority scheme to service requests.
- FIG. 10B shows queue processor state machine 162 , which couples to transmit arbiter 80 , block address and byte count registers 164 in control memory, and transmit queue 136 .
- FIG. 11 shows DMA engine 104 , which couples to CPU register 166 , CPU master interface 36 , XQMB 154 , packet memory interface 42 , and control memory interface 160 .
- DMA engine handles data transfer between packet memory 16 and CPU local memory 100 so that CPU 12 may perform other tasks.
- CPU 12 packet send is enabled by creating packet in local memory 100 , register set-up, and initiating packet transfer.
- packet receive is enabled by notifying CPU 12 .
- CPU 12 checks block attribute to determine whether to process packet. If CPU 12 transfers packet to local memory 100 , it DMA engine 104 is notified to proceed. Otherwise, register is written to de-queue packet.
- FIG. 12 flow chart shows CPU 12 to packet memory 16 operation. Initially, in software, CPU sets-up register and initiate packet transfer 168 . Then, in hardware, processor determines 170 whether to initialize block attribute 176 , whether 172 to initialize DMA transfer 178 , and whether 174 to write command to XQMB 180 .
- FIG. 13 flow chart shows packet memory 16 to CPU 12 operation. Initially, in hardware, block attribute is read 182 , and CPU 12 is notified 184 . Then, in software, run CPU check whether DMA needed 186 ; if not, then set register to de-queue 188 . Then, in hardware, DMA transfer 190 , and notify CPU 12 .
- FIG. 14 shows switch circuit 20 with L 3 engine 70 coupled to FE 52 , interface 36 to CPU 12 , interface 42 to packet memory 16 , IP header RAM 198 , MAC address RAM 196 , interface 194 to L 3 CAM 126 , and interface 160 to control memory 136 .
- L 3 check block captures destination IP address, Time To Live Field and Checksum field in L 3 Header Memory, for use by L 3 block 70 for L 3 lookup and processing.
- L 3 check block processes rest of packet header. Receive packet is checked for IP protocol filed, and to detect packets for specialized handling.
- IP header length is checked to determine whether packet needs specialized option processing. If header length is not equal to 5 32-bit words, option processing is applied to packet. Time To Live filed is checked to see if TTL field is more than 1; if less than 1, packet is marked with TTL error flag. IP packet length is checked for minimum length to contain full IP header. Checksum of header is performed.
- L 3 INFO Memory is 32-byte wide. Each location is dedicated for corresponding numbered port. Result of L 3 header checks for receiving port is stored in corresponding location and used by Forwarding Block to decide whether packet is sent to L 3 Block for processing.
- L 3 check (e.g., CHK) block takes into consideration if arriving packet contains VLAN tag, if VLAN tag option is enabled. If so, hardware accounts for shift in appropriate fields for L 3 header checking process. This amounts to 4-byte shift of L 3 header following MAC header.
- VLAN priority bits are extracted and passed along with L 3 INFO. VLAN priority bits may be enabled to override QoS information set in L 2 CAM result and L 3 Header Lookup result. Programmable register is provided to load pattern to identify if incoming packet is VLAN tagged packet.
- L 3 Engine (L 3 E) 73 is hardware block for implementing the Layer 3 CAM lookup and age table functions for switch circuit 20 .
- L 3 E 72 receives requests from forwarding engine (FE) 52 and CPU 12 , processes requests and returns results to requester.
- L 3 E 72 lookup functions includes: receiving, buffering and processing lookup requests from FE 52 , providing hardware to calculate hash index from destination IP (DstIP) address provided by FE 52 ; reading CAM entry at address and checking for IP address match; following linked list entries until match is found or end of list is reached; and returns lookup result to FE 52 .
- DstIP destination IP
- L 3 E 72 age table maintenance function includes: maintaining age table in control memory 136 ; adding and deleting entries in table by CPU 12 request; aging table at CPU-controlled intervals; reporting aged entries to CPU; maintaining aging time stamp; and making entries live.
- L 3 E 72 CAM management assistance function includes: providing hardware hash calculation function for CPU 12 ; implementing search function which scans L 3 CAM and reports matching entries; and providing change option to search function which writes new data into matching entries.
- CPU 12 interface to L 3 Engine 70 is for age table and L 3 CAM maintenance.
- Initial CAM entries are written to L 3 CAM 126 by CPU 12 through dedicated control memory interface port. Managing linked entries and free buffers is done by CPU 12 . Searching for entries and reporting or changing them is accomplished by appropriate command registers.
- Age table entries are created and deleted by CPU 12 using add and delete commands. Aged entries are reported to CPU 12 and deleted by CPU 12 using delete command. Time hardware modifies age table entry when entry is made.
- Packet memory 16 includes 8-MB SDRAM with 4 1M ⁇ 16 devices providing 32-bit data path to 4096 2KB blocks for packet storage.
- L 3 Engine 70 writes to packet memory 16 to modify fields (e.g., destination address (DA), source address (SA), TTL and checksum) in packet following L 3 lookup.
- DA and SA fields are written in 32-byte burst with byte enables set appropriately.
- MAC address RAM 196 is 32-entry RAM, indexed by port number, which contains lower byte of MAC address for each physical port.
- IP HDR RAM 198 is 2-port Internet Protocol header memory RAM located on switch circuit 20 . Each entry contains IP values (e.g., TTL, checksum and DST IP) for packet. Write port of RAM 198 is used by packet memory engine 90 to store data from packet IP header. As data streams to packet memory 16 , appropriate bytes are pulled and written to RAM 198 . L 3 Engine 70 uses read port of RAM 198 to access data required to process lookup request from FE 52 . Entries are indexed by port number, so receive (RCV) port number is used to lookup entry.
- IP values e.g., TTL, checksum and DST IP
- Write port of RAM 198 is used by packet memory engine 90 to store data from packet IP header. As data streams to packet memory 16 , appropriate bytes are pulled and written to RAM 198 .
- L 3 Engine 70 uses read port of RAM 198 to access data required to process lookup request from FE 52 . Entries are indexed by port number, so receive (RCV) port number is used to lookup entry.
- L 3 CAM 126 is contained in 2-MBytes synchronous DRAM (SDRAM) located in single 1M ⁇ 16 part. Since SDRAM is optimized for burst transfer, L 3 Engine 70 accesses occur in bursts of eight 16-bit words. On-chip arbiter/controller logic for L 3 CAM 126 memory has multiple ports to allow better pipelining of accesses and L 3 engine 70 uses two of these ports.
- SDRAM 2-MBytes synchronous DRAM
- L 3 CAM 126 data structure is implemented as hash table combined with pool of free buffers which can be linked to entry in hash table. Entry, whether in hash table or free buffer pool, is 8 words (16 bytes). Entry is referred to by entry number, 17-bit number used when indexing CAM, when indexing into age table or when reporting results of search or aging operation.
- Base hash table contains 64K entries and resides in lower 1-MByte SDRAM. Entries in table have entry numbers in 0 to 64K range, i.e. bit 16 of entry number is set to ‘0’. Entries located in free buffer pool are in upper 1-Mbyte of SDRAM, and entry numbers have bit 16 set to ‘1’. Address of first word of entry in CAM is determined by concatenating entry number with 3 bits of ‘0’.
- CPU 12 creates entries in hash table for DstIP addresses by hashing address and using resulting 16-bit hash index as offset to entry in table. When multiple address' hash to same entry in base table, link is created to free buffer pool entry. If additional addresses hash to same location, they can be added to end of linked list. CPU 12 creates and maintains entries and manages linked list structures.
- Control memory block (CTL MEM) 136 uses 128K ⁇ 16 synchronous SRAM (SSRAM), instead of SDRAM devices because most data structures stored require single read and write accesses.
- L 3 Engine 70 uses 32-KB portion of control memory to store age table. It does single read followed by single write of word in age table. Each 16-bit word contains age table information for 4 CAM entries. Aging information for particular L 3 CAM 126 entry is accessed by using CAM entry number divided by 4 as address into age table.
- Forwarding Engine (FE) 52 performs lookup requests to L 3 Engine 70 for each IP packet to be processed. Four-deep FIFO buffer is provided to buffer these requests. FE 52 provides RCV Port Number and Block Number for each packet. After lookup is complete, L 3 Engine 70 returns RCV Port Number as well as L 3 Result and L 3 Status word containing various flags and information from matching CAM entry.
- age table support since control memory 136 containing age table does not support locked operations, table modifications are done by hardware. Such table modifications address condition of two agents trying to modify same table entry.
- CPU 12 can initialize entries to invalid state at startup by writing to control memory; but during operation, hardware performs table modifications.
- Age table operations are done by CPU 12 write to age command register.
- Write to age command register causes Age Table Busy flag in L 3 Status register to be set until operation is complete.
- Aged entries are reported in registers (e.g., AgeResult 1&2).
- age table maintenance is illustrated, starting with CPU or live command processing 200 , then determine if age command 202 applies. If so, increment time stamp 210 and set age flag; otherwise, read table entry 204 , mask and modify table entry 206 , and write table entry 208 . Further, in FIG. 15B, after age flag set 212 , age table is read 216 , then determine age out 218 . If so, then write result registers 220 , set result valid 222 , wait for CPU 224 , and clear result valid 226 ; otherwise determine 228 if last entry. Next, clear age flag 232 , 230 and read hash table 240 .
- Time stamp is 2-bit value providing four age time intervals. There are two age time counters, currTime and ageTime. CurrTime is reset to zero and increments when CPU 12 issues age command. Entries with time stamps equal to this value are newest entries. AgeTime value is always equal to currTime +1 m(i.e., currTime ⁇ 3, modulo 4). Entries with time stamps equal to ageTime are aged next time CPU 12 issues age command.
- CPU adds entry to age table when creating new entry in L 3 CAM 126 . Until entry is added to age table, entry does not participate in aging process.
- CPU 12 writes (e.g., AgeCmd) register with entry number and add or add permanent command, and hardware reads appropriate entry, modifies valid and permanent bits appropriately and writes currTime into time stamp field.
- Hardware makes entry live (i.e., accessed) when L 3 CAM lookup results in IP hit. Entry number of matching entry is used to access age table, and time stamp field is updated with currTime. Entries which are accessed frequently have more recent time stamp than infrequently used entries, and are not aged out.
- CPU 12 deletes entry in age table when removing entry from L 3 CAM 126 .
- CPU 12 writes AgeCmd register with entry number and delete command, and hardware reads appropriate entry, clears valid bit, and writes modified entry back to table.
- First aged entry number used to access L 3 CAM 126 to retrieve DstIP for entry.
- DstIP is hashed to locate base hash table entry and CAM entry at address is read.
- Hardware follows linked list, reading CAM entries until retrieving entry with Link Address equal to original aged entry number. Entry number is reported along with aged entry number in AgeResult registers.
- CPU 12 provides L 3 CAM management functions, including initial setup, adding entries, deleting entries and managing linked lists and free buffer pool.
- Hardware provides automatic search/change capability to assist CPU 12 in locating entries with certain characteristics and optionally changing such entries.
- Search operations are initiated by CPU 12 write to SearchCmd register.
- Write to SearchCmd register causes Search Busy flag in L 3 Status register to set until operation is complete.
- Matching entries are reported in (e.g., SearchResult) registers.
- FIG. 16A shows search operation steps.
- CPU initiates search 234 , writes commands 236 , initialize entry to zero 238 , read hash table 240 , then determine match 242 . If so, write result to registers 244 , and wait for CPU 246 ; otherwise, determine if linked 248 . If so, clear age flag 250 , else, determine if last entry 254 . If so, clear age flag 252 ; otherwise, clear age flag 256 .
- Hardware performs automatic and exhaustive search of L 3 CAM 126 when SearchCmd register is written. Starting with entry 0, each entry in base hash table is read and checked against search criteria. If entries have valid link address, then linked entries are read and checked. Minimum 64K CAM entries are read.
- SearchCmd can be written with Abort Flag set, and hardware exits search process. Pending SearchResults are read by CPU 12 before hardware exists and clears Search Busy flag.
- SearchMask and SearchData registers (16 registers total). Before search command is issued, SearchMask registers are written. ‘0’ in bit position masks bit from consideration in comparison. SearchData registers are written with data values to be matched. Registers containing data to be matched are written.
- change option was selected when (e.g., SearchCmd) value was written, then matching entries found during search are changed by hardware according to values written to change setup registers. When matching entry is found, hardware alters data and writes back to CAM before reporting match result to CPU 12 .
- NewCamDataX ChangeMaskX & ChangeDataX
- L 3 Engine receives CAM lookup requests from forwarding engine and searches matching entry in L 3 CAM 126 . Results of search are returned to FE 52 , and additional requests are serviced.
- FIG. 16B flow chart shows CAM lookup steps. Initially, valid buffer is set 258 , read IP header RAM 260 , hash destination IP (DstIP) address 262 , and read hash table 264 , then determine if valid and hit 266 . If so, read port MAC RAM 268 , update packet data 270 , modify packet 272 , and write result 274 . If not, determine valid link 276 . If so, follow link 278 and read hash table 264 , else write result 274 .
- DstIP hash destination IP
- L 3 Engine 70 buffers up to 4 lookup requests from forwarding engine (FE) 52 . When buffer is full, busy signal is sent to FE 52 . Buffer is organized as FIFO and contains receiving port number and block number for lookup request.
- DstIP address is hashed to 16-bit value which is used as entry number for base hash table. That entry is read, and words containing DstIP address are compared to packets DstIP address. If these two addresses match, then IP hit bit is set, and results of successful lookup are returned to FE 52 .
- packet Before result is posted to FE 52 , packet may be modified, depending on bit in L 3 Flags field of CAM entry. If Don't modify bit of CAM entry is set, nothing is changed in packet. Otherwise, when lookup is successful, TTL field of IP header is decremented and modified in packet memory, and (e.g., CheckSum) field is recalculated and changed. Packets DA is overwritten with value contained in matching CAM entry, and SA is replaced with value from MAC Address Registers and MAC Address RAM.
- L 3 Flags field of CAM entry If Don't modify bit of CAM entry is set, nothing is changed in packet. Otherwise, when lookup is successful, TTL field of IP header is decremented and modified in packet memory, and (e.g., CheckSum) field is recalculated and changed. Packets DA is overwritten with value contained in matching CAM entry, and SA is replaced with value from MAC Address Registers and MAC Address RAM.
- L 3 CAM 126 If match is not found in L 3 CAM 126 , hardware checks to see if default route registers are written by CPU. These registers provide ‘default route’ CAM entry and are programmed with same information as CAM entries in control memory 136 . If default route exists, then packet is modified using default information, and (e.g., IPHit and DefaultRouteUsed) bits of L 3 Result are set.
- default route registers are written by CPU. These registers provide ‘default route’ CAM entry and are programmed with same information as CAM entries in control memory 136 . If default route exists, then packet is modified using default information, and (e.g., IPHit and DefaultRouteUsed) bits of L 3 Result are set.
- MAC Address RAM 196 contains lower byte of MAC address for each port. It is 32 ⁇ 8 dual port ram which is written by CPU 12 and read by hardware during packet modification. This value replaces lower byte from MAC Address Registers when writing new SA for packet.
- L 3 results of CAM lookup returned to FE include receive port number and block number originally provided by FE 52 and two 16-bit values, L 3 Result and L 3 Status. A detailed bit definition for these last two values was provided earlier in this document.
- Switch circuit 20 operates on various performance and integrity levels. In cut-thru switching mode, relatively fast ether switching mode and high performance level are achieved; however, there is possibility of transmitting packet with error during receive process. In this mode, CPU programs MACs to raise receive request when collecting 64 bytes. Also it programs MAC to raise subsequent receive request after every 64 bytes collection. First request provides fast header analysis and switching.
- SF store forward
- Switch circuit 20 waits until packet completion and updates transmit queues of relevant destination. MAC programming remains same.
- Forwarding Block acts on ‘store forward mode’ mode on port by port basis. SF mode is selectable as per port basis. In this mode port linking is disabled.
- packet moves in following directions: received on LAN port(x) and transmitted to any/all other LAN ports; received on LAN port(x) and posted to UL queue; received on LAN port(x) and posted to CPU queue; received on LAN port(x) and packet dropped; and forwarded from CPU to any/all LAN ports.
- packet flows through packet memory 16 and switch circuit 20 .
- switch circuit 20 participates in forwarding, filtering and queue management.
- switch circuit 20 In case of Ethernet port originated packet flow, packet is received on Ethernet ports, and switch circuit 20 is triggered on such packet from request from one of MACs. This is hardware trigger mode. Switch circuit 20 , in coordination with RISC processor 12 , allocates free block pulled from receive free list. Once block is assigned, block is busy until destination agent(s) complete transmission. Transmission completion has mechanism to release block and insert in receive free list.
- switch circuit 20 needs to obtain packet header information. Header is extracted from MAC received data stream.
- PM engine 90 identifies header from data stream and loads on port-specific segment of Ether header memory.
- CAM Processor makes lookup on CAM and delivers result to Auto Forwarding Block.
- AFB adds packet to one of following queues: one of Ethernet ports transmit queue; all Ethernet ports queues, UL transmit queue and CPU queue; UL queue; CPU queue; or L 3 block for L 3 lookup.
- AFB handles updating “Block address”, “Byte count”, and “routing information” on transmit queues. Once such information is provided, respective transmitting agents handle packet transmission. At end of transmission, block is released and added to receive free list. Refer for more details about releasing the block in some other sections.
- CPU 12 posts packets to XQMB 154 of Auto Forwarding Block for transmission to one or several ports.
- XQMB 154 handles posting packet to respective queues.
- CPU 12 assembles or modifies existing packet in packet memory 16 for transmission.
- CPU 12 with help of DMA function, can transfer packet from local memory 100 to packet memory 16 , and at end of such transfer, can initiate XQMB 154 action.
- FIG. 17 shows packet receive process, which is accomplished by Receive Arbiter 80 , Buffer scheduler 94 , LAN bus controller 76 , Packet memory engine 90 , header memory 290 , data buffer 84 , as well as Auto Forwarding Block (AFB) and PM SDRAM controller.
- Receive Arbiter 80 receives packet receive process
- Buffer scheduler 94 receives packet data from packet data buffer 84 .
- Packet memory engine 90 Packet memory engine 90
- header memory 290 e.g., data buffer 84
- data buffer 84 et data buffer
- ARB Auto Forwarding Block
- PM SDRAM controller PM SDRAM controller
- Receive arbiter 80 arbitrates and prioritizes receive request from Ether ports. It raises request to buffer scheduler 94 . When request is under process, arbiter 80 makes background processing on remaining requests.
- Buffer scheduler 94 handles resource allocation of internal buffers.
- Buffer scheduler 94 maintains two receive buffers and two transmit buffers 84 .
- Each buffer can hold up to 64 bytes of data, and optimizes buffer allocation algorithm for fair bandwidth extraction between receivers/transmitters.
- LAN Bus Controller 76 interfaces to LAN bus to read/write packet data to MAC FIFO buffers, and access MAC receive completion status.
- LAN bus controller 76 may access MAC and read data slice, and store to internal data buffer.
- a data buffer can hold up to 64 bytes of information.
- Packet memory engine 90 sets-up moving packet in slice-by-slice into PM 16 .
- Packet memory engine 90 reads PM receive block register and byte count register, and updates (i.e., increments) byte count register on each transfer.
- Packet memory engine 90 commands PM SDRAM controller to start data transfer.
- PM SDRAM controller transfers data from receive buffer to packet memory 16 , and generates control timing to access external SDRAM.
- Auto Forwarding Block allocates free block to receiver; initializes receive block register, byte count register. Occasionally AFB commands to reject packet.
- Every LAN MAC has (e.g., RREQx) receive request signal which, when active, including at least 64 bytes of data (i.e., header/data region) is collected in internal FIFO. There are 32 request signals from LAN bus. Following steps describe new packet reception and header memory loading:
- RREQx signal becomes active, indicating 64 bytes valid in FIFO.
- Request active and (e.g., RBODYx) bit clear means new packet.
- RREQx signal is first-level conditioned if corresponding bit enabled in Receive Enable register.
- RISC 12 allocates free block to receive port, it writes block address on corresponding receive block register and enables the receiver”.
- Conditional RREQx first wins RREQ arbitration to get service.
- Buffer scheduler 94 allocates one of two free receive data buffer and enables LAN bus controller 76 to start data transfer.
- LAN bus controller 76 executes Burst Read Accesses on LAN bus targeted to Port-x. Read data is written on allocated internal receive buffer. Since body bit is clear, loading process signatures slice as “header”. If slice is header, it writes header data on 2-port Ether Header memory. At end of header loading, Port to be Analyzed FIFO is loaded with 5-bit port number. Loading of FIFO enables CAM engine to start analyzing header information. Load completion calls attention of PM engine for data movement from receive data buffer to packet memory 16 .
- PM engine 90 updates byte count and sets-up SDRAM Controller for data transfer to packet memory 16 .
- Buffers have dedicated channels to SDRAM Controller.
- SDRAM Controller arbitrates transfer requests amongst channels and starts executing request at time overlapping address and data phases to maximize throughput and efficiency. Requesting channel is held arbitrating for LAN bus until full slice is moved in packet memory 16 .
- Packet data reception procedure similarly to MAC RREQx signal which, when active informs at least 64 bytes data, is collected in internal FIFO, as follows:
- RREQx signal is first-level conditioned if corresponding bit is enabled in Receive Enable register. Conditional RREQx first wins the RREQ SCAN arbitration to get the service.
- Buffer scheduler 94 allocates free receive data buffer 84 and alerts LAN bus controller 76 .
- LAN bus controller 76 executes burst-read accesses on LAN bus targeted to Port-x. Read data is written in allocated receive buffer 84 .
- Rec Link register(x) indicates accordingly. If link bit is set, slice is in switch mode for present slice and consecutive slices until end of packet.
- Loaded data buffer calls attention of PM engine 90 to load data into packet memory 16 .
- PM engine 90 uses Receive Block Address register(x) and Receive BC register(x) to construct PM destination address. Byte count is updated in receive BC register(x) and transmit BC register(y).
- PM engine issues command to PM SDRAM controller to start data transfer and ready to service receive or transmit buffer or accept command.
- Forwarding Engine in coordination with CAM Processor and L 3 Lookup Block, evaluates current receiving packet for following possible decisions: reject packet; link packet; forward packet to transmitter queue; multicast packet to two or more ports; broadcast packet only to Ether ports; broadcast packet to Ether ports and UL; send packet to UL; send packet to CPU; or send packet to L 3 Lookup Block for L 3 analysis.
- CAM Processor writes decision information into header analyzed FIFO. Such write process wakes-up Forwarding Block to take-up forwarding process.
- PM engine 90 keeps loading successive slices of packet in packet memory 16 independent of CAM analysis. Decision of CAM might occur at middle of packet reception or after end of packet reception. If header analysis is complete before packet reception is complete, Forwarding Block acts on packet if packet is unicast or destination is L 3 Lookup Block which carries on further analysis associated with L 3 forwarding. For other cases, Forwarding Block is not called to action until receive completion of packet. If packet reception is complete before header analysis, Forwarding Block is not called into action until header analysis is complete. Receiver is not primed again until forwarding decision has been taken on received block and acted upon.
- control bit is set for corresponding receiving port to reject incoming packet.
- PM engine 90 looks at reject bit to set prepare for transfer from receive buffers to packet memory. If reject is set, PM engine 90 empties FIFO without setting transfer to PM. PM engine 90 clears reject bit at end of packet reception. Receive complete state is indicated to Forwarding Block.
- CAM lookup posts port number in CAM Analysis Done FIFO in addition to setting CAM analysis done bit for port. This draws attention of Forwarding Block prior to completion of packet reception. Forwarding Block checks several conditions to take forwarding action. At this time, it may link packet to corresponding transmitter or post packet in queue of transmitter.
- Transmitter may be busy, i.e., transmitter queue contains one or more packets queued or transmitter is currently processing old packet.
- Forwarding Block requests XQMB 154 to post receive packet in transmitter queue with incomplete information. This is handled by manipulating RC bit clear in BC entry in control memory 136 . Bit, if clear, means packet block address is valid, but byte count is invalid. Packet data is incomplete in packet memory 16 .
- Forwarding Block pushes incomplete packet in transmitter queue on special occasion. When, receiver(x) wants to switch to transmitter and if transmitter is currently busy, Forwarding Block puts packet in transmitter queue to maintain order of priority. At pushing event, byte count information is invalid.
- XQMB 154 commands to link. If receiver completes packet before getting to transmitter, Forwarding Block sets such bit, and loads valid BC value on BC entry. Subsequent forwarding action on packet degenerates to store and forward mode.
- Transmitter may be free when switching decision occurs.
- Forwarding Block commands to link receiver to transmitter. It does not manipulate control memory structure.
- Forwarding Block primes receiver, and transmitter continues to transmit until end of packet without further intervention. Finishing transmitter event releases block and pushes to receive free list.
- Forwarding Block may act on receive packet after receive completion and CAM analysis completion. Since packet is received, XQMB 154 is instructed to post on appropriate transmitter queue; this is Store and Forward mode.
- Broadcasting to Ether ports decision may result from not finding destination port or hit on broadcast MAC address.
- broadcast map for receiving port is fetched, and packet is forwarded to transmit queue management block for posting on transmit queues
- header analysis results in port not physically connected to LAN ports, but CPU or uplink ports; Forwarding Engine Block instructs XQMB 154 to queue on appropriate ports.
- XQMB 154 may queue packet on ports if ports are specified in broadcast port map.
- LAN controller When end of packet is sensed from LAN port(x), LAN controller signals by bit in slice status. PM engine 90 , while moving slice to PM 16 , notifies same status by setting appropriate bit in Rec end reg. Forwarding Block acts on every receive completion; and in addition to forwarding actions, it instructs Free Queue Management Block to prime receiver. In case previous packet is rejected, no new block needs to be allocated; in such case it enables receiver to receive new packet.
- transmit port activity is top-level enabled by Forwarding Block or Transmit Queue Management Block of Auto Forwarding Block.
- XQMB 154 picks highest priority packet, and loads transmit block address register and byte count register corresponding to packet. This action enables transmitter on transmit enable register.
- Forwarding Block or XQMB 154 loads link command with which hardware copies receiver block address to transmitter block address register. It copies current running receiver rec(x) byte count value to transmitter(y) byte count register. It also sets link bit active.
- Transmitter enters arbitration if transmit (e.g., XMT) enable bit is set, and Byte count validity is met. If MAC transmit FIFO has at least 64 bytes free space, it raises TREQ# signal. This signal is conditioned with first phase enable signal, and transmitter enters arbitration with other TREQ# signals asserted by other transmitters. Winning transmitter requests allocation of one of two free transmit buffers. This request is forwarded to buffer scheduler. When buffer scheduler allocates free buffer, transmitter enters arbitration for PM engine 90 service. PM engine 90 time-multiplexes between receive requests and transmit requests and other commands such as link, receive enable and transmit enable.
- transmit e.g., XMT
- PM engine 90 time-multiplexes between receive requests and transmit requests and other commands such as link, receive enable and transmit enable.
- PM engine 90 moves sets-up transfer with PM SDRAM controller by giving command to move slice from PM to data buffer, and updates byte count and address registers in array for corresponding transmitter.
- PM engine 90 signatures slice as header or non-header based on XMT body bit. Along with slice, PM engine 90 passes information, such as slice count and port address through buffer attributes. Loaded slice calls attention of corresponding LAN controller for service to transfer data from transmit buffer to MAC on LAN bus. LAN controller moves slice to target MAC port and releases buffer. Whenever PM engine 90 moves slice, decremented byte count is compared, to check, if reached zero. If reached zero, packet may reach end of packet status based on following cases:
- transmitter byte count reaching zero is not regarded as end of packet; it is regarded as transmitter has to wait for receiver to get slice. Transmitter does not participate in arbitration again until slice is received on linked receiver. Link bit clear and byte count zero signal packet completion.
- PM Engine clears XMT enable bit, and sets End of packet transmit bit. End of transmit draws attention of XQMB 154 to look at transmitter queue in control memory 136 . If queue contains additional packets, XQMB 154 loads new packet to re-enable transmitter. If queue is empty, XQMB 154 does not take action. Trigger point for enabling transmitter is: when current packet ends, and new packet is pending in queue; when receive packet is targeted to transmitter and queue is empty; or when CPU inserts packet to transmitter.
Abstract
Description
- Invention relates to digital networks, particularly to multi-layer switching network apparatus and method.
- Conventional local area network (LAN) and TCP/IP have become dominant technologies in computer networking. As businesses increasingly rely on such technologies, both LAN size and TCP/IP traffic volume that runs across them have grown dramatically. This has led the network manager on continuous search for products to increase network performance, easily adapt to changing network requirements, and preserve existing network investment.
- Presently, LAN technology is evolving into Gigabit per second (Gbps) range. Equipment designers have been challenged to make network interfaces and networking products such as bridges, routers, and switches, fast enough to take advantage of the new performance. Compounding the equipment design problem has been the rapid innovation in networking protocols. The traditional response to this shifting sands problem has been to build easily upgradable software-intensive products. Unfortunately, these software intensive products typically exhibit poor system performance.
- Accordingly, there is need for a new generation of internetworking devices capable of gigabit speeds, but with the flexibility of previous software intensive products.
- Invention resides in a multilayer switching device and associated technique for enabling simultaneous wire-speed routing at layer3, wire-speed switching at
layer 2, and support multiple interfaces atlayer 1, according to OSI reference model. Inventive implementation may be embodied using one or more integrated circuits (ASIC), RISC processor, and software, thereby providing wire-speed performance on interfaces, in various operational modes. - FIG. 1 is system-level diagram of preferred embodiment.
- FIGS.2A-B are block diagrams of first- and second-level switch respectively of present embodiment.
- FIG. 3 is general switch block diagram of present embodiment.
- FIG. 4 is general control-path diagram of present embodiment.
- FIG. 5 is general datapath diagram of present embodiment.
- FIGS.6A-B are block diagrams of LAN interface and datapath interface respectively of present embodiment.
- FIGS.7A-B are block diagrams of DMA transfer between local memory and packet memory, and processor access to packet memory respectively of present embodiment.
- FIGS.8A-B are processor access to L3CAM memory and control memory respectively of present embodiment.
- FIGS.9A-B are processor access to L2CAM memory, and LAN arbiter interaction with datapath respectively of present embodiment.
- FIGS.10A-B are transmit queue management block (XQMB) interfaces and operation respectively of present embodiment.
- FIG. 11 is DMA block diagram of present embodiment.
- FIG. 12 is flowchart of CPU-to-packet memory operation of present embodiment.
- FIG. 13 is flowchart of packet memory-to-CPU operation of present embodiment.
- FIG. 14 is block diagram of L3 block interfaces of present embodiment.
- FIGS.15A-B are flowcharts of age table maintenance of present embodiment.
- FIGS.16A-B are flowcharts of search and lookup operations respectively of present embodiment.
- FIG. 17 is flowchart of packet reception of present embodiment.
- FIG. 1 is top-level overview diagram of system architecture for preferred embodiment. Multilayer switch device6 couples local area network (LAN)
workgroup hubs 2 throughenterprise switching hub 4 to wide-area network (WAN) links through multiprotocol router 8. Multilayer switch 6 and associated technique enables simultaneous wire-speed routing at Layer 3 (L3), wire-speed switching at Layer 2 (L2), and support multiple interfaces at Layer 1 (L1), according to OSI reference model. System may be embodied using one or more integrated circuits (ASIC), RISC processor, and software, thereby providing wire-speed performance on various interfaces in various operational modes. - System architecture comprises two-level distributed multilayer switch, preferably using 4-Gbps non-blocking switch fabric6. Multilayer (i.e., both L2 and L3) switch fabric is entirely contained within single ASIC capable of switching 3M pps or more. A 4 Gbps I/O bus connects one or more interface modules to the ASIC. Because the switch matrix is not necessarily integrated with the MAC layer, a wide range of interface types can be supported (i.e., both LAN and WAN). Using present embodiment, various combinations of
layer 1 interfaces are supportable, and all interface modules are field-upgradable. Various interface modules may carry multiple physical interfaces. - As shown in FIG. 2A, first-
level switch 22 includes switch ASIC 20, which couples RISC coprocessors (i.e., Network Management Processor (NMP) 10 and Route/Switch (RS)processor 12,) for supporting for higher-layer software functions and support features. Optional components may be added for redundancy of critical system components, such as power supplies.Memory 16 and input/output (I/O)modules 14 couple to switchcircuit 20. - In FIG. 2B, second-level switch or
cross-bar interconnection 18 couples multiple first-level switches 22. For example, in configuration shown with six first-level switches 22, aggregate performance of non-blocking switch fabric may exceed 24 Gbps. -
RISC processors switch element 22 execute software to provide standards-based dynamic routing, and non-real time activities such as network management. Software is stored in flash memory, and is network-updatable via TFTP. Preferred software functions include: dynamic Internet Protocol (IP) routing (e.g., RIP, RIPv2, OSPF);layer 2 support (e.g., 802. ID STP); configuration support (e.g., enable/disableLayer 2 or Layer 3 support on per-port basis; ports can be grouped into broadcast domains, flexible subnet configuration); network management: (e.g., SNMP, HTML, Telnet, TFTP, DHCP support). - Additional software functions include: quality-of-service provisioning (QOS) (e.g., providing multiple levels of prioritization, address- and policy-based QOS, dynamic layer3, QOS based on RSVP); IP Multicast (e.g., IGMP, DVMRP); network traffic monitoring (e.g., RMON); hot standby support (e.g., VRRP); additional dynamic routing (e.g., NHRP); and certain IEEE enhancements (e.g., 802.IQ (i.e., VLAN), 802.3x (i.e., flow control), and 802.1p (i.e., priority)).
- Present multi-layer switch approach offloads multiprotocol router8 of local IP routing, thereby leaving router 8 with bandwidth for routing other protocols, and for handling WAN links. Hence, existing investment in router 8 may be preserved, for example, without changes to WAN topology. Further, effective performance at network apex is wire speed, and
enterprise switching hubs 4 at network apex may be segmented, thereby preserving bandwidth, and extending useful life. Additionally, with present system architecture, installation therein of network products and applications is comparatively easier, particularly because addressing changes are incremental, thereby minimizing impact on network operations. Moreover, preferred system not use non-standard protocols, thereby assuring interoperability in multi-vendor environment. - Although present multilayer switch system is suitable for applications at network aggregation points, present system may also be used in high-performance workgroup and server applications. For example, in high-performance workgroup application, present system may interconnect between cluster of closely cooperating high performance computers, such as in video postproduction, where ability to transfer data rapidly between workstations is critical to production throughput. In such case, wire-speed performance is interesting, and flexible layer3-addressing support provides connections outside workgroup, without impacting switching speed. Additionally, in case of server applications, present multilayer switch system provides network attachment point for one or more servers. Wire-speed performance of present system allows network designer to use either
layer 2 or layer 3 topologies, and removes potential network performance bottleneck. - Moreover, as described in further detail hereunder, preferred implementation of innovative multilayer switch apparatus and methodology provides following functionality: support for 16 or more full-duplex 100BaseT ports or up to 28 ports of 10/100BaseT ports; direct interface to MIPS-type RISC processor for management and routing; integration of SDRAM controller for shared high-speed 6-channel packet memory; integrates of CAM access interface to system processor; integration of hardware CAM processor for L2 learning, lookup and live interactive activities or transactions; integration of hardware hash-based IP header lookup and management; integration of hardware-based transmit and free queue management; integration of L2 and L3 forwarding of unicast, broadcast and multicast packets; broadcast traffic management; integration of QoS, with 4 priority queues per port; hardware-handled packet movement; integration of 768 bytes of dual-port memory for L2 and L3 header for 28 ports; support for 4MB/16MB of SDRAM packet memory; implementation of 256 bytes of data buffers for concurrent transfers to PM SDRAM and LAN bus; intelligent buffer scheduler & arbiter for efficient bandwidth distribution; low-latency mode, store and forward mode selection, with 10-us switching latency; operation of LAN bus at 64-bit/66-Mhz; operation of packet memory bus at 32-bit/100-Mhz; operation of processor bus at 32-bit/66-Mhz; operation of control memory and L2 CAM interfaces at 16-bit/66-Mhz; operation of router (RT) cache SDRAM interface at 16-bit/66-Mhz.
- Preferably,
multilayer switch circuit 20 is implemented as single-chip integrated circuit (e.g., semicustom ASIC) for processing handles switching of any canonical packet, cell, frame, or other data communication element, with no or limited processing assistance from external processors.Switch circuit 20 operates in relatively low latency, and store-and-forward switching modes. Transactions between Ethernet ports may operate in low-latency cut-thru mode; other transactions may occur in store-and-forward mode. - As appropriate,
switch circuit 20 may contain substantially one or more of following functions: external bus interface, processor interface, CAM interface, LAN interface, packet memory (PM) SDRAM interface, route cache SDRAM interface, control memory (CM) SRAM interface, LAN block, LAN bus arbiter, LAN bus controller, LAN block interfaces, data path block, data path buffers, data path controller, buffer scheduler, packet memory, packet memory SDRAM arbiter and controller, DMA function-to-processor interface, packet engine (PE), port control function, port attribute memory, L2 CAM engine, memory blocks for header and CAM analysis result, CAM structures, L2 header analysis hardware engine, auto-forwarding block, forwarding block, L3 header analysis result memory, free queue management block, block attributes management, transmit queue management block (XQMB), SRAM arbiter and controller, processor interface, L3 block, L3 header memory, hash function, L3 lookup algorithm, L3 management function, L3 aging function, route cache (RC) SDRAM arbiter and controller, RISC processor interface, slave interface, bus master interface, DMA interface, bus protocol, register interface-to-internal resources, and interrupts. - In FIG. 3, for example, preferred implementation of switch system shows general logic block diagram for
switch circuit 20 coupled to: 64-bit 66 Mhz LAN bus,external memory 16 through 32-bit 99-Mhz bus, L2 CAM through 16-bit 66-Mhz bus,control memory 136 through 16-bit 66-Mhz bus, L3 route cache through 16-bit 66-Mhz bus, and switchprocessor 12 through 16-bit 66-Mhz bus, which couples to network management processor (NMW) 10 through external interprocessor controller (IPC) 24. - In FIG. 4, internal control path of
switch circuit 20 is shown.External switch processor 12 couples to CAM interface 46, free queue management 48, L3 lookup 50, transmit queue management and scheduler 58,SDRAM memory controller 62, andSRAM memory controller 64. Also, internal control path includes forwardingengine 52, which couples to CAM interface 46, free queue management 48, L3 lookup 50, block attributes 60, transmit queue management and scheduler 58, and receive block 54. Transmit queue management and scheduler 58 couples to transmit block 56,SRAM memory controller 64, and block attributes 60. Receive block 54 and transmit block 56 couple to LAN bus. CAM interface 46 couple to CAM bus and receive block 54.SRAM memory controller 64 couples to free queue management 48, block attributes 60, L3 lookup 50, andSDRAM memory controller 62.SDRAM memory controller 62 couples to RC memory bus and L3 lookup 50. Block attributes 60 couples to free queue management 48. Forwardingengine 52 couples to receive block 54. - In FIG. 5, internal datapath of
switch circuit 20 is shown. In particular, multi-channel packet memory arbiter andcontroller 66 couples to SDRAM packet memory bus, processor andDMA interface 68,L3 engine 70, receiver buffers 72, and transmitbuffers 74. Receive and transmitbuffers 72, 74 couple to media access controller (MAC) first-in first-out (FIFO) bus. Processor andDMA interface 68 couples to processor bus. -
Switch circuit 20 includesprocessor interface 36 which couples to 32-bit MIPS RISC processor multiplexed bus (e.g., NEC R4300). Such processor bus, a 32-bit address/data bus operable up to 66 Mhz, operates in master and slave modes. In slave mode, such processor bus responds to accesses to internal resources, such as registers,CAM 142,Control Memory 136, PM SDRAM and RC SDRAM. In master mode, such bus handles DMA operations to and from PM SDRAM. Such processor bus does not respond to accesses to external resources, but cooperates with external system controller circuit. In master mode, such processor bus may handle DMA to system memory. -
Switch circuit 20 includes CAM interface 46, a dedicated 16-bit bus compliant with content-addressable memory (i.e., Music Semiconductor CAM 1480 compatible) operating at 66 Mhz. Such bus may be shared by external interface. For route/switch (RS) processor accesses to CAM memory, special data path is provided throughswitch circuit 20.Switch circuit 20 generates CAM access timing control on behalf ofRS processor 12.Switch circuit 20 learns and looks-up MAC addresses and port numbers through such bus. -
Switch circuit 20 includesLAN interface 40 which couples LAN bus, a 64-bit access bus operating at 66 Mhz. Ethernet MAC devices connect to such LAN bus through receive and transmit MAC FIFO bus.Switch circuit 20 generates select signals and control signals for access to external MAC device FIFO bus.Switch circuit 20 reads writes data in 64-bit single-cycle burst mode. Burst size is 64 bytes. Preferred bandwidth is 4 GB/s at 64-bit/66 Mhz-operation at 64-byte slice size. Ethernet frames are transferred across LAN bus. At end of receive frame, status bytes are read. -
Switch circuit 20 includes packet memory (PM)SDRAM interface 42, which includes PM SDRAM bus which operates at 32-bit/99-Mhz standard.Packet memory 16 is directly connected to such bus through registered transceivers. Preferred bandwidth is 400 MB/s at 99-Mhz operation and 64-byte burst mode. Seven-channel arbiter insideswitch circuit 20 allows up to 7 agents to accesspacket memory 16. PM interface supports up to 8 MB of SDRAM in two banks. -
Switch circuit 20 includes interface to Route Cache (RC) SDRAM for coupling timing control signals and multiplexed 16-bit bus, which operates in 66-Mhz mode capable of streaming data at 132 MB/sec. -
Switch circuit 20 includes interface to Control Memory (CM) SRAM for managing block free queue list, transmit queues, block parameters and L3 CAM aging information. Such interface is 16-bits wide and operates at 66-Mhz. Address and data buses are multiplexed and operate in flow-through and pipelined modes. - FIG. 6A shows LAN block and interfaces40 externally to Ethernet Media Access Controller (MAC) FIFO bus and internally to CAM interface block,
datapath block 44, andpacket engine block 82. LAN block interface functionality include bus arbitration for receive and transmit requests of FIFO bus, bus control and protocol handling, signaling internal datapath block to initiate data transfers and communicating with packet engine to signal begin and end of receive and transmit operations on FIFO bus. As shown,datapath block 44 couples to FIFO data bus, LAN bus controller 76,buffer allocator 78, andpacket engine 82. LAN bus controller (LBC) 76 couples to FIFO bus control,buffer allocator 78, and receiver and transmitarbiters 80, which couple topacket engine 82 and receive and transmit requests. - When LAN interface40 operates, receive requests and transmit requests are multiplexed and fed by external logic. Multiplexer uses 2-bit counter output. Front end demultiplexer reconstructs requests on 32-bit receive request register and 32-bit transmit request register. Few clocks latency for request may be sensed to be activated or deactivated, which may be handled by
arbiter mechanism 80. - Receive
arbiter 80 services receive port requests, preferably in round-robin scheme for equal distribution. Overlapped processing provides improved performance. Hence, if receive port is under service, next request prioritization occurs in parallel. During arbitration,arbiter 80 may receive port enabled, free block allocated signals from other modules. Upon certain channel winning arbitration, internal receive buffer is allocated 78, and data staged from MAC FIFO bus forpacket memory 16. When buffer is granted, channel is presented to LAN Bus controller 76 for data transfer. - Additionally, transmit
arbiter 80 services transmit port requests in round-robin scheme for equal distribution. Overlapped processing provides improved performance. Hence, when transmit port is under service, next request is prioritized in pipeline. During arbitration,arbiter 80 may receive port enabled, valid packet assigned, in link mode the transmitter has at least one slice signals from other modules. If channel has data slice indatapath 44, channel is not allowed to join arbitration until data is put intopacket memory 16, thereby preventing out-of-sequence data transfer. Upon channel winning arbitration, it is presented tobuffer allocator block 78 to obtain internal transmit buffers for staging frompacket memory 16 for MAC FIFO bus. Once transmit request wins arbitration, and transmit buffer is allocated, channel is presented topacket engine block 82 to obtain data frompacket memory 16. Once data is staged in transmit buffer, buffer requests to LAN Bus controller 76 to transfer data in transmit buffer to MAC FIFO bus. - LAN bus controller76 provides access to MAC FIFO bus targeted to port moving slice between MAC FIFO and internal data buffers. Receive request, which wins receive arbitration and secures one of receive buffers from
buffer allocator 78 and transmit buffers having data for transfer to FIFO bus, competes for services of LAN bus controller 76. Arbitration mechanism is configured to split bandwidth evenly between receive requests and transmit requests. LAN bus controller 76 generates end-of-packet status read cycles for receive request data transfer operations. Status information is used to determine if received packet is good or bad. If error is sensed, received packet may be rejected. - Data bus width of LAN bus is 64 bits. LAN bus access is performed in burst mode (i.e., single-cycle burst mode) with maximum of 64-byte transfer, preferably executing at 8 data cycles in burst. LAN bus controller76 is started by buffer scheduler when data buffer is allocated to receive or when data transfer from
packet memory 16 to one of transmit buffers is complete. - Receive and transmit data to LAN bus is staged through 64-byte deep receive and transmit data buffers in
datapath block 44. Receive and transmit requests arbitration and FIFO bus control are handled by LAN block.Buffer allocator 78 indatapath block 44 manages allocation of receive and transmit buffers, andpacket engine block 82 handles movement of data betweenpacket memory 16 and receive and transmit buffers. - FIG. 6B shows
datapath block 44 interface, includingpacket memory controller 82 coupled to data buffers 84 and packet memory engine (PME) 90. Data buffers 84 couple to LAN block 86,buffer scheduler 94, slice counters 88. Buffer attributes 92 couple toPME 90 andLAN block 86, which couple to bufferscheduler 94. - Data transfers between packet memory bus and MAC FIFO bus are staged through receive and transmit
buffers 84 indatapath block 44. Block logic tracks state ofbuffers 84.Datapath block 44 interacts withLAN Block 86,packet engine block 82 andpacket memory controller 82. - Data transfers between MAC FIFO bus-to-
PM 16 andPM 16 to MAC FIFO bus occur through temporary datapath storage buffers 84 insideswitch circuit 20.Buffers 84 match difference in bus access bandwidth for slice, and maintain concurrent transfers between FIFO bus and PM bus. - Two buffers are provided for transmission, and two buffers are provided for reception. Such buffers are associated with respective buffer status. Transmit buffers hold data form
PM 16 to MAC FIFO (LAN) bus. Receive buffers hold data from MAC FIFO bus toPM 16. Each buffer has dedicated channel to PM SDRAM Controller. PM SDRAM Controller arbitrates each request to transfer on first-come/first-serve basis. On LAN side, appropriate buffer is selected for read or write. - Frame transfer across LAN bus occurs on slice basis. Slice is 64 bytes. When
switch circuit 20 is servicing port, slice of data transfers on single-cycle burst mode. Burst data transfer size is slice size, except for last slice in frame. Last slice size is decided by frame size. Ports are serviced, in time-division multiplex mode. - Receive slice buffer is used to capture LAN data from MAC FIFO. Slice is 64 bytes.
Switch circuit 20 has two 64-byte buffers. During LAN FIFO read access, incoming 64-bit data words are strobed on selected slice buffers, word-by-word, during clock edges. Write order is from top to down. Receive status is maintained for respective receive slice. For example, slice status provides: - Receive slice size (represented by 6-bit number.) Maximum is 64 bytes. In read access, MAC provides in each data phase, valid bytes through bits (e.g., LBE#<7-0>). Hence, LBEI#<7-0> are registered and analyzed at end of data phase to provide cumulative slice size.
- EOF signaling. MAC provides in each read data phase, if end-of-frame. EOFI# signal is registered and stored for EOF status. It is also used to close current transfer.
- SOF signaling. MAC provides on each read data phase, if Start-of-frame. SOFI# signal is registered and stored for SOF status.
- Transmit slice buffer is used to capture (e.g., PMDO) bus data and supply to LAN bus. Slice is 64-bytes. Switch circuit has two 64-byte slice buffers. During LAN FIFO write access, 64-bit data words are read from selected slice buffer. One clock pre-read is implemented to provide minimum delay time on LAN data (LD) bus. Read order is from top to down.
- Status is maintained for respective transmit slice. Slice status is loaded by
PM engine 90 when moving slice from PM. Status information includes: - Slice size (represented by 6-bit number.) Maximum is 64 bytes. When slice is read from PM bus, PM engine registers slice size.
- EOF signaling.
PM engine 90 registers signal while transferring slice from PM bus. If status is on, LAN FIFO controller asserts EOF# signal at appropriate data phase. - SOF signal. PM engine registers signal while transferring first slice of packet from PM. If status is on, LAN FIFO controller asserts the SOF# signal at first data phase.
-
Buffer scheduler 94 allocates transmit and receive data buffers to requesting agents, keeps track of busy/free status of each buffer, and allocates free buffer to requesting agent.Buffer scheduler 94 optimizes for (a) equal distribution of bandwidth between receivers and transmitters, (b) avoiding deadlock situation of transmit buffer, and (c) achieving highest concurrence of LAN bus and PM bus. - Datapath controller includes buffer attributes92 for receive and transmit
buffers 84, and track byte count per slice basis. Buffer attributes 92, such as End-of-Packet (EOF), start-of-packet (SOF), Byte Enables (BEB), and Slice Count are tracked from time data arrives into receive or transmit buffer until data leaves buffer.Buffer attribute 92 information is used bypacket memory engine 90 to track progress of packet flowing throughswitch circuit 20 per slice basis. Datapath controller interacts withbuffer scheduler 94 at end of slice transfer to release buffer. Synchronization between PM SDRAM controller andLAN bus interface 40 is thereby accomplished. - Packet memory resides on dedicated SDRAM bus.
Switch circuit 20 integrates SDRAM controller to accesspacket memory 16. PM SDRAM controller functionality includes: 32-bit interface operating at 99-Mhz to 8 MB of external SDRAM; support for up to 7 internal requesting agents; arbitrates requests and generates request to SDRAM control block; pipelines requests for maximum efficiency and throughput; bursts of 4 (one bank), 8 or 16 (both banks) accesses on SDRAM; and maximum performance at 16 bursts and minimum performance at single read or write. - Route processing is provided by MIPS R4000
family RISC processor 12, which interfaces with switch circuit through address/data multiplexed bus. RISC processor interface may use external system controller, for example, for communicating withswitch circuit 20 though processor slave port. RISC processor serves switch orroute processor 12. Several register resources inswitch circuit 20 are used byRISC processor 12 to control configuration and operation ofswitch circuit 20.RISC processor 20 may access resources outside ofswitch circuit 20, such access being controlled byswitch circuit 20packet memory 16, route cache memory, and CAM for L2 forwarding.Switch circuit 20 communicates status of operation and draws attention ofprocessor 12 through status and process attention registers. When configured,switch circuit 20 performs DMA of data from packet memory to processor local memory, and forwards packets to processor queue maintained byswitch circuit 20. - Preferably, route processor (RP)12 is NEC Vr4300 RISC microprocessor from MIPS family with internal operating frequency of 133 Mhz and system bus frequency of 66 Mhz.
Processor 12 has 32-bit address/data multiplexed bus, 5-bit command bus for processor requests and data identification, six handshake signals for communication with external agents, and five interrupts. Bus width can be selected as 32-bit operation.Processor 12supports Processor 12 uses little endian when accessing switch resources. -
RP 12 is interfaced to switchcircuit 20.RP 12 communicates withNMP 10 through interprocessor communication (IPC)bus 24, and accesses switch local resources, such aspacket memory 16, L3 CAM (Route Cache) 28,control memory 136 andL2 CAM 142 throughswitch circuit 20 and local resources, such as local memory, ROM etc., through system controller. Two interrupts are used byswitch circuit 20 to issue interrupt requests toprocessor 12. Two slaves onRP processor 12 system bus are switch and system controller. Switch is final agent to provide ready signal to processor requests that switch or system controller is ready to accept. During DMA transfer, switch acts as master. - Write access is implemented as ‘dump and run’ with two pipelined buffers to improve system performance. This allows two back-to-back write cycles. One read request is processed at a time.
Processor 12 accesses internal registers resources in 32-bit mode. Write buffer and read buffer are provided topacket memory 16 to match frequency difference of 99-Mhz and 66-Mhz. Memory interface to switch is 32-bit. Maximum burst size topacket memory 16 is four 32-bit words (i.e., 16 bytes). Read buffers are provided to L3CAM andcontrol memory 136 because of 16-bit interface to switch. Little endian is used when data is packed and unpacked during write or read requests to 16-bit interfaced memories. Maximum burst size to L3CAM 28 is 16 bytes, and to CM is 8 bytes. Write or read request to memories is arbitrated through agents inside switch, such as forwarding engine, L3 engine etc., so latency depends on various factors. - During write access,
processor 12 owns mastership or control of bus. During read requests,processor 12 enters into uncompelled slave state after address phase, giving bus control to external agent to drive data. - FIG. 7A illustrates DMA transfer between
RP processor 12local memory 100 andpacket memory 16. DMA transfer betweenpacket memory 16 and NMP processor local memory is also provided in architecture. NMP processor system controller responds to DMA master requests between packet memory and NMP processor local memory. DMA is implemented using two design blocks calledDMA engine 104 and DMA master 102. DMA engine is interfaced topacket memory 16 and that of DMA Master to processor system bus. DMA is initiated by setting bits in DMA command register. DMA transfer betweenlocal memory 100 andpacket memory 16, or vice versa, occurs substantially as follows: -
DMA engine 104 notifies DMA master 102 to initiate DMA transfer when packet is pending by giving request. DMA master 102 arbitrates for processor bus withRP processor 12 as another master by giving request (e.g., EREQ) toprocessor 12. During DMA transfer,switch circuit 20 acts as master tosystem controller 98. Processor gives bus control to RP processor when ready. When bus is granted by processor, DMA transfer begins. Mastership of processor bus can be re-acquired byRP processor 12 between each slice transfer, which is maximum of eight 32-bit words (i.e, 32 bytes).DMA engine 104 reasserts request after each slice transfer, until block of packet data is transferred. At end of DMA, bus control is given to processor. - When bus is in uncompelled slave state, DMA master102 does not access processor system bus to simplify design. While DMA transfer is taking place on bus,
system controller 98 does not drive bus, assuming bus in slave state. - FIG. 7B illustrates
RP processor 12 access to packet memory (PM) 16 through L2/L3 switch circuit 20. Switch interface topacket memory 16 is 32-bit, and maximum burst size is 16 bytes. Synchronous DRAM is chosen for packet memory that can be operated at 66-Mhz, 99-Mhz and 125-Mhz frequencies. During processor write request, processor dumps write-data into front-end pipeline buffers 106.Slave state machine 108 provides such data into packetmemory write buffer 110. Processor request is arbitrated with LAN requests and L3 engine requests in PM SDRAM arbiter to accessPM 16.PM SDRAM controller 112 generates control signals for SDRAM. During processor read request, read-data is provided in PM read buffer 114 from packet memory bus.Synchronizer 116 converts 99-Mhz signal into 66-Mhz pulse signal that initiates slave state machine to empty read buffer. Read data is muxed with data from other blocks and driven to processor system bus. Packet memory to local memory (PM-to-LM) DMA transfer data is not written into read buffer, but passed to processor system bus. - FIG. 8A illustrates
RP processor 12 access to L3CAM (route cache)memory 28.RP processor 12 accesses L3CAM 28 throughswitch circuit 20 to initialize entries and maintain data structures. Additionally, FIG. 8B showscontrol memory 136 access throughswitch circuit 20. For both such memory accesses in FIGS. 8A-B,RP processor 12 couples to switchcircuit 20 through 66-Mhz, 32-bit processor system bus, whereinpipeline buffers 106 receive processor write data and couple toslave state machine 108. - In
switch circuit 20 shown in FIG. 8A, L3CAM write buffer couples toslave machine 108 andL3CAM SDRAM controller 120, which receive requests from other agents and couples toL3CAM memory 28 through 66-Mhz, 16-bit bus. L3CAM readbuffer 122 provides read data through 32-bit processor bus and couples toslave state machine 108 and register 134 over 16-bit bus.Register 134 receives 66-Mhz clock signal and couples toL3CAM memory 28 through 66-Mhz, 16-bit bus. - In
switch circuit 20 shown in FIG. 8B, CM write buffer 128 couples toslave machine 108 andCM SSRAM controller 130, which receive requests from other agents and couples to controlmemory 136 through 66-Mhz, 16-bit bus. CM read buffer 132 provides read data through 32-bit processor bus and couples toslave state machine 108 and register 124 over 16-bit bus.Register 124 receives 66-Mhz clock signal and couples to controlmemory 136 through 66-Mhz, 16-bit bus. - Synchronous SDRAM is chosen for
L3CAM 28, and Synchronous SRAM is chosen forcontrol memory 136. Switch interface to both memories is 16-bit, and both memories operate at 66-Mhz.Processor 12 access to memories is similar in both cases, maximum burst size to L3CAM memory is 16 bytes, and maximum burst size forcontrol memory 136 is 8 bytes. Data is packed and unpacked for each processor access. - Each
memory pipeline buffers 106 byslave state machine 108. Since memory interface is 16-bit, processor write data is divided into two 16-bit half words.Processor 12 request toL3CAM memory 28 is arbitrated with L3 engine requests.Processor 12 request to control memory is arbitrated with forwarding engine, FQMB, L3 engine, XQMB, BAM and DMA. During processor read request, L3CAM or CM read data from memory bus is provided in read buffer. When last transfer is triggered,slave state machine 108 starts emptyingread buffer 122 appropriately and packs two half words into 32-bit word, and puts on processor system bus. - In FIG. 9A,
RP processor 12 accessesL2CAM memory 142 throughswitch circuit 20. Content Addressable Memory (CAM) is chosen for accessingL2CAM memory 142, which operates at 66-Mhz frequency.Switch circuit 20 interface to L2CAM memory is 16-bit.Processor 12 executes commands write/read and data write/read to L2CAM 142 using CAM access registers provided insideswitch circuit 20.Processor 12 accessesL2CAM 142 through register-based request-grant handshake by loading L2CAM Access Control & Status Register to execute processing cycles. -
RP processor 12 arbitrates withCAM arbiter 138 inswitch circuit 20 for CAM bus. For processor requests,slave state machine 108 generates control signals forCAM 142, andCAM arbiter engine 138 processes switch requests. During processor write request,processor 12 provides write-data in pipeline buffers 106. When CAM bus is granted byCAM arbiter 138,slave state machine 108 puts data frompipeline buffer 106 on CAM bus. During read request, read-data from CAM bus is muxed 140 with data from other blocks and passed to processor system bus. Write/read buffers need not be provided in present case. - FIG. 9B shows LAN arbiters interaction with datapath. Register files144 for receivers and transmitters, including corresponding block address registers and byte count registers, couple to block address register and byte count register coupled to
state machines 148, which couple to switchdata path 44 and receive and transmitarbiter 150. -
Packet switch engine 82 performs control functions for transfer request in and out of receive and transmit buffer topacket memory 16.Packet engine 82 handles LAN data movement, command process, and PM address calculation. - For LAN data movement,
packet memory engine 82 sets up for moving slice betweenpacket memory 16 and allocated data buffer. This is triggered by scheduler while slice is scheduled to move in/out of data buffer. PM engine has access right to block address registers and registers to understand actual address onPM 16 and update packet size. - For command process,
packet memory engine 82 executes systematic hardware processes when Forwarding Block and Transmit Queue Management Block (FB/TQMB) generates instructions such as: link, receive enable, transmit enable, receive reject, etc. Hence, end of packet reception/transmission is noticed for next packet initialization. In notifying such events, priority encoding is provided for first-in/first-service integrity. - For PM address calculation,
packet memory engine 82 regards Ethernet ports as 32 concurrent full-duplex DMA channels. Relevant PM pointers for each channel are maintained. For every slice transfer, PM bus address is calculated. - Preferably, buffer attributes92 are provided in attribute block address array, which is 3-port architecture having 64×12-bit 3-port memory array. Port-i is write port; port-2 is read port; and port-3 is read port.
Packet memory engine 82 can write/read memory locations using various ports. Forwarding engine (FE) can read locations asynchronously. Port-3 is assigned for FE. - First 32 locations are used for “Receive block address” of 32 receive ports. Next 32 locations are “Transmit block address” for 32 transmit ports.
PM Engine 82 initializes block address for receive/transmit ports on command of Auto Forwarding Block.PM 10engine 82 reads block address relevant to receive/transmit port under service.PM engine 82 uses block address to identify packet inPM 16. - CAM interface block analyzes incoming packet at
layer 2, i.e., at MAC layer. Analysis result is forwarded to Auto Forwarding Block state machine. CAM processor is called for attention when ether header block is loaded in ether header memory. On such trigger condition, after acquiring CAM bus interface, CAM Processor starts defined fast processing action. Block containslayer 2 header memory, analyzed and to be analyzed port FIFOs, and result memory. CAM block interfaces to internal memories organized as memories and FIFOs as well as external CAM to accomplish L2 lookup. - When begin transfer of receive data, begin of header is identified and required header information is loaded into ether header memory. Sixteen-byte header blocks are reserved for each port in header memory. Loaded indication is updated on 5-bit entry in 32-deep ether header to-be-analyzed FIFO. Such FIFO provides first-in/first-service feature.
- Ether header memory is 2-port memory has 64×64 bit architecture. Port- I is write port, and port-2 is read port. Such memory is located on LAN side of receive buffer. As first slice of new receive packet is loaded into receive buffer, header slice (i.e., 16 bytes) is written to ether header memory in 64-bit words. Ether port number is used as reference address to select header block number. Maximum of 32 header blocks can be stored in such memory. Port-2 is used by CAM processing engine. CAM engine reads 16-bit quantity at a time through front-end 64:16 multiplexer. L3 header information, up to 8 bytes per port is stored in different memory. Such information is used by L3 lookup engine during routing operation.
- Ether-to-be-analyzed FIFO memory is 32×5-bit two-port memory, holding maximum of32 port numbers to be analyzed. Port-1 is write port, and port-2 is read port. FIFO is written with port number when first slice of data is received on LAN bus and header loaded in ether header memory. CAM Processor reads port number through port-2 for indexing header memory. FIFO structure ensures that ports to be analyzed are presented to CAM engine in arrived order.
- Ether analyzed FIFO memory is 32×6-bit two-port memory, holding
maximum 32 analyzed port numbers. Port-1 is write port, and port-2 is read port. CAM Processor writes analyzed port number through port-1 and Forwarding Engine (FE) reads through port-2. FIFO structure ensures that analyzed ports are presented to forwarding engine in arrived order. - Ether result memory is 32×16-bit two-port memory, holding results for 32 ether ports. Port-1 is write port, and port-2 is read port. CAM Processor writes L2 forwarding result through port-1, and Forwarding Block (FB) reads through port-2. When CAM Processor processes specific ether port header, it uses port number as address to write result. FB reads port number from Analyzed FIFO to make forwarding decision. FB uses port number as reference address to read CAM analysis result.
- External CAM memory is 1024×64-bit capacity on standard configuration. Size can be expanded to 2048×64-bit by adding CAM device in vertical expansion. CAM memory is connected on dedicated CAM bus. Such bus is shared between CPU and switch circuit. Normally such bus is default-owned by switch circuit. CPU can use bus by register mode bus handshake.
- CAM memory contains1024 locations of 64 bits wide. Locations can be assigned as RAM property or CAM property. Location assigned as RAM will not be accounted in lookup process; CAM locations participate in lookup process. Repeatedly used parameters are stored in RAM so that real-time data movement between RAM location and Comparand/mask registers/etc. can happen with minimum overhead. Every location has associated status filed which describes entry, such as: empty entry, valid entry, skip entry on lookup, and RAM entry.
-
Layer 2 header analysis is performed by CAM processor. Ethernet headers are loaded and processed on dedicated Ethernet header memory having 128×32 bit dual port memory. Assuming case where packet received on port(x),switch circuit 20 is triggered on such packet by request from MAC port number(x), which is effectively hardware packet arrival notification. - Header is extracted from MAC received data stream. New receive packet data is identified with arrival of SOF, and first 16 bytes are treated as layer-2 header. If header is concurrent to store access to receive buffer, then header is stored in port-specific block number(x) in header memory. Writing process may not account for port contention. Block written on port-1 may not be accessed on port-2. Header is stored as header block(x). At end of storage, port number is written in ether-to-be-analyzed FIFO, which is 32×5-bit register. FIFO write pointer is incremented after each status write
- CAM processor starts when valid entry is loaded in Ether-to-be-analyzed FIFO. CAM Processor maintains read pointer to read valid entry. Valid entry is notified if there is difference between write pointer and read pointer. Entry read provides port number of header. CAM Processor uses port number to reach header block(x).
- Preferably, switch system has 32 ports and 32 entries. New packet on port can not be received unless old packet is processed, according to system-level handshake. Hence, at any time, no more than 32 headers/header status may be stored, effectively reducing complexity of FIFO design. PM engine can blindly write header/status without looking for FIFO-full condition. CAM Processor can start as long as pointers are not equal.
- CAM processor handles header processing. CAM processor is notified of Ethernet header valid when write pointer and read pointer differ. When entry is valid on Ethernet-to-be-analyzed FIFO, CAM processor reads entry and increments read pointer. Using such value, CAM processor can reach specified header block. Ether header memory is divided into 32 blocks. Port number directly provides starting address of header block. Entries in block are consecutive 16 bytes.
- CAM processor processes header block, and writes result on port specific location on Ether result memory. CAM process completion is notified to Auto Forwarding Block through Ethernet result FIFO, which is 32 deep register construction. Each entry is 6-bit wide. Entry is result of CAM memory lookup. If set, destination MAC address indicates CAM hit. Routing tag in header block is valid. If clear, CAM lookup fails; routing tag does not contain valid information.
- To write on result FIFO, CAM processor has write pointer, which is 5-bit counter. CAM processor write entry, whereas AFB read entry. When CAM completes process, it writes result entry, and increments write pointer. At last, CAM processor increments Ether header status FIFO read pointer to point to next entry.
- CAM processor header processing includes learning process of: source lookup, source port read, and source learning. CAM processor learns MAC addresses arriving from Ethernet ports. As associated process of CAM lookup, CAM processor determines whether source address was learned previously, i.e., by reading source address from Ether header memory, and writes CAM for lookup. If match occurs, processor presumes source port was learned; it reads existing port information from associated data to compare whether port is same as receiving port. If MAC header matches, whether or not ports match, processor makes entry live and at same time relearns receiving port. If receiving port number does not match learned port, Source Address (SA) Learned flag is set. If miss, processor learns entry into next free address if CAM is not full, and if learned SA Learned flag is set. While updating such new entry, processor follows correct data structure for RAM associated information.
- Optionally, attribute is set with (e.g., ETHR_LRN_INHIBIT) register for each port to inhibit learning on specified ports. If set, during source lookup process, after source port read, entry is made live if hit and if miss, MAC address is not learned. Source port read phase can be skipped if source port filtering is not required.
- Destination lookup process includes steps: destination lookup and destination port read. CAM processor reads 6-bytes destination MAC address from header memory and writes on CAM for comparison lookup. If miss, destination is assumed unknown; if hit, destination is available through memory (e.g., ARAM) field, which provides destination port number and port/MAC address specific status and control flags. If hit, CAM processor reads ARAM field and writes in result memory, setting hit flag. If miss, CAM processor has nothing to read and write miss flag to result memory. Rest of result data is not valid in miss case. Forwarding blocks read this field for analysis and forwarding decision. At end of process, CAM analysis done flag is set for packet on receiving port.
- CAM processor analyzes results of source lookup and destination lookup processes to determine how to process incoming packet. Each port has two bits allocated to handle spanning-tree protocol requirements. One bit is allocated for ‘Port Blocked State’ flag and other for ‘Learn Inhibit’ flag, which is used for learning of MAC addresses on receiving port. Port Blocked State’ flag is used for forwarding decision. Filtering bits in result from both source lookup and destination lookup along with port specific STP control bits relating to forwarding, source port and destination port read as result of destination lookup and read are considered.
- CAM processor sets CAM analysis completion status for receiving port. If destination lookup resulted in hit and destination port is one of physical LAN ports and cut-thru switching on port is enabled or CPU port, port number is written to Ether analyzed FIFO. CPU port is allowed to enable Layer3 analysis parallel to packet reception. Result processing is done by Auto Forwarding Block (AFB). AFB is notified of CAM process completion through Ethernet Analyzed FIFO. AFB can read highest priority FIFO entry using hardware hidden read pointer. If read pointer and write pointer are different, one or more valid entries are available in Ethernet analyzed FIFO. AFB reads valid entry and gets port number. Reading entry increments read pointer, if present entry is valid.
- Using port number, AFB can access Ether result memory. Refer to Auto Forwarding Block section for details on AFB functionality. If CAM analysis resulted in miss or hit but packet can not be switched, AFB does not need to be notified until packet reception is complete. Hence, CAM processor merely sets CAM analysis completion flag for receiving port. AFB processes packet when both receive completion and CAM analysis completion set for receiving packet.
- Aging process is performed by
processor 12 as processor bandwidth requirement for task is relatively low. Time stamp register provides variable granularity for aging. Processor uses instruction set provided by CAM device. Entries to be aged are processed in one instruction, though setup is required before executing instruction. In addition to status bits provided by CAM for every entry, 3 bits in RAM field are dedicated for aging information. Status provide by CAM are used to identify if entry is ‘Valid’, ‘Empty’, ‘Skip’ or ‘RAM only’. One of bits allocated in ARAM field is used to mark entry ‘Permanent’. Entries marked ‘Valid’ and not ‘Permanent’ are considered for aging. Additional two bits in ARAM filed provide flexibility to CPU to implement aging process. - When entry is visited during source lookup process of CAM analysis, if source is found, ARAM filed is updated with latest time stamp from (e.g., ETHER CAM CONTROL) register. If new source is learned, in addition to port, time stamp bits are written into ARAM field. When processor visits CAM to ageout entries, it searches CAM for entries with oldest time stamp. In search process, processor configures mask registers in CAM in such way that age bits enter comparison, and entries that are not ‘Valid’ and marked ‘Permanent’ do not enter comparison. In next instruction, processor can clear ‘Valid’ bits on matching locations to ‘Empty’ state. By doing so, oldest entries are marked empty. From that point, aged entries do not enter compare operation until made ‘Valid’ aging during normal learning process.
- Auto forwarding block (AFB) is hardware Ethernet packet forwarding engine and queue processor. AFB analyzes incoming packet and may forward packet both at
layer 2 and layer 3. After forwarding analysis is done, AFB posts and maintains port queues. AFB may accept packets from processor interface and post packet in requested queues. AFB provides processing power on packet-by-packet and manages required information for integrity of packet routing strategy. AFB feeds initial setup information for each ether packet for each port to run data transaction. - AFB functionality enables
switch circuit 20 to perform forwarding and filtering without real-time assistance fromprocessor 12. Processing element is out of the datapath, and forwarding and filtering is done at line rate for supported ports. AFB functionality includes: free queue management, block attributes management, receive port management, forwarding and filtering, transmit queue management, quality-of-service (QoS) support, and control memory interface. - Forwarding function features port linking, wherein receive port is linked to transmit port before packet is fully received, thereby improving latency between received and transmitted packets. Port linking is accomplished in forwarding stage if conditions are suitable. For example, packet can cut-thru with unique destination, i.e. no more than one port is target destination for packet. Cut-thru enable bit are satisfied for certain values, such as: destination port, speed-matching logic, xmtPortEn, xmtPortNotBsy, xmtQNotVld and mirrNotVld. Data arrive speed should not exceed transmitting port speed. Transmitter should be ready to accept command. Transmitter may be busy transmitting data, or there may be packets waiting for queue transmit Also, there should be minimum of data present in buffer before process can start or arbitration latency may result in transmit FIFO under-run condition. In such case, transmitter is linked but does not start transmitting data until required minimum data is received in packet memory.
- FIG. 10A shows Transmit Queue Management Block (XQMB)154, which is hardware block for managing transmit queue functions for
switch circuit 20.XQMB 154 couples to forwarding engine (FE) 52,DMA Engine 104, block attribute memory (BAM) 152,PME 90, queue attribute memory (e.g., AttrRAM) 156,port AttrRAM 158, andcontrol memory 136 throughinterface 160.XQMB 154 functionality includes: initializing and managing transmit queues for each port; maintaining QOS parameters (i.e., tokens) for each port; queues (e.g., nQueues and dQueues) blocks to/from control memory transmitter queues; forwarding blocks to requesting transmitter; returning block numbers toBAM controller 152; forwarding multi/broadcast block in ‘background’; supporting 28 physical ports, 3 logical ports and multi/broadcast port; and using round-robin priority scheme to service requests. Furthermore, FIG. 10B shows queueprocessor state machine 162, which couples to transmitarbiter 80, block address and byte count registers 164 in control memory, and transmitqueue 136. - FIG. 11
shows DMA engine 104, which couples toCPU register 166,CPU master interface 36,XQMB 154,packet memory interface 42, and controlmemory interface 160. Generally, DMA engine handles data transfer betweenpacket memory 16 and CPUlocal memory 100 so thatCPU 12 may perform other tasks.CPU 12 packet send is enabled by creating packet inlocal memory 100, register set-up, and initiating packet transfer. Also, packet receive is enabled by notifyingCPU 12.CPU 12 checks block attribute to determine whether to process packet. IfCPU 12 transfers packet tolocal memory 100, itDMA engine 104 is notified to proceed. Otherwise, register is written to de-queue packet. - FIG. 12 flow chart shows
CPU 12 topacket memory 16 operation. Initially, in software, CPU sets-up register and initiatepacket transfer 168. Then, in hardware, processor determines 170 whether to initializeblock attribute 176, whether 172 to initializeDMA transfer 178, and whether 174 to write command toXQMB 180. - FIG. 13 flow chart shows
packet memory 16 toCPU 12 operation. Initially, in hardware, block attribute is read 182, andCPU 12 is notified 184. Then, in software, run CPU check whether DMA needed 186; if not, then set register tode-queue 188. Then, in hardware,DMA transfer 190, and notifyCPU 12. - FIG. 14 shows switch
circuit 20 withL3 engine 70 coupled toFE 52,interface 36 toCPU 12,interface 42 topacket memory 16,IP header RAM 198,MAC address RAM 196,interface 194 toL3 CAM 126, andinterface 160 to controlmemory 136. - As packet is received, L3 check block captures destination IP address, Time To Live Field and Checksum field in L3 Header Memory, for use by
L3 block 70 for L3 lookup and processing. L3 check block processes rest of packet header. Receive packet is checked for IP protocol filed, and to detect packets for specialized handling. - IP header length is checked to determine whether packet needs specialized option processing. If header length is not equal to 5 32-bit words, option processing is applied to packet. Time To Live filed is checked to see if TTL field is more than 1; if less than 1, packet is marked with TTL error flag. IP packet length is checked for minimum length to contain full IP header. Checksum of header is performed.
- Results of above checks are written into informational (e.g., L3 INFO) Memory. L3 INFO Memory is 32-byte wide. Each location is dedicated for corresponding numbered port. Result of L3 header checks for receiving port is stored in corresponding location and used by Forwarding Block to decide whether packet is sent to L3 Block for processing.
- L3 check (e.g., CHK) block takes into consideration if arriving packet contains VLAN tag, if VLAN tag option is enabled. If so, hardware accounts for shift in appropriate fields for L3 header checking process. This amounts to 4-byte shift of L3 header following MAC header. Optionally, VLAN priority bits are extracted and passed along with L3 INFO. VLAN priority bits may be enabled to override QoS information set in L2 CAM result and L3 Header Lookup result. Programmable register is provided to load pattern to identify if incoming packet is VLAN tagged packet.
- L3 Engine (L3E) 73 is hardware block for implementing the Layer 3 CAM lookup and age table functions for
switch circuit 20. L3E 72 receives requests from forwarding engine (FE) 52 andCPU 12, processes requests and returns results to requester. L3E 72 lookup functions includes: receiving, buffering and processing lookup requests fromFE 52, providing hardware to calculate hash index from destination IP (DstIP) address provided byFE 52; reading CAM entry at address and checking for IP address match; following linked list entries until match is found or end of list is reached; and returns lookup result toFE 52. L3E 72 age table maintenance function includes: maintaining age table incontrol memory 136; adding and deleting entries in table byCPU 12 request; aging table at CPU-controlled intervals; reporting aged entries to CPU; maintaining aging time stamp; and making entries live. L3E 72 CAM management assistance function includes: providing hardware hash calculation function forCPU 12; implementing search function which scans L3 CAM and reports matching entries; and providing change option to search function which writes new data into matching entries. -
CPU 12 interface toL3 Engine 70 is for age table and L3 CAM maintenance. Initial CAM entries are written toL3 CAM 126 byCPU 12 through dedicated control memory interface port. Managing linked entries and free buffers is done byCPU 12. Searching for entries and reporting or changing them is accomplished by appropriate command registers. - Age table entries are created and deleted by
CPU 12 using add and delete commands. Aged entries are reported toCPU 12 and deleted byCPU 12 using delete command. Time hardware modifies age table entry when entry is made. -
Packet memory 16 includes 8-MB SDRAM with 4 1M×16 devices providing 32-bit data path to 4096 2KB blocks for packet storage.L3 Engine 70 writes topacket memory 16 to modify fields (e.g., destination address (DA), source address (SA), TTL and checksum) in packet following L3 lookup. DA and SA fields are written in 32-byte burst with byte enables set appropriately.MAC address RAM 196 is 32-entry RAM, indexed by port number, which contains lower byte of MAC address for each physical port. -
IP HDR RAM 198 is 2-port Internet Protocol header memory RAM located onswitch circuit 20. Each entry contains IP values (e.g., TTL, checksum and DST IP) for packet. Write port ofRAM 198 is used bypacket memory engine 90 to store data from packet IP header. As data streams topacket memory 16, appropriate bytes are pulled and written toRAM 198.L3 Engine 70 uses read port ofRAM 198 to access data required to process lookup request fromFE 52. Entries are indexed by port number, so receive (RCV) port number is used to lookup entry. -
L3 CAM 126 is contained in 2-MBytes synchronous DRAM (SDRAM) located in single 1M×16 part. Since SDRAM is optimized for burst transfer,L3 Engine 70 accesses occur in bursts of eight 16-bit words. On-chip arbiter/controller logic forL3 CAM 126 memory has multiple ports to allow better pipelining of accesses andL3 engine 70 uses two of these ports. -
L3 CAM 126 data structure is implemented as hash table combined with pool of free buffers which can be linked to entry in hash table. Entry, whether in hash table or free buffer pool, is 8 words (16 bytes). Entry is referred to by entry number, 17-bit number used when indexing CAM, when indexing into age table or when reporting results of search or aging operation. - Base hash table contains 64K entries and resides in lower 1-MByte SDRAM. Entries in table have entry numbers in 0 to 64K range, i.e.
bit 16 of entry number is set to ‘0’. Entries located in free buffer pool are in upper 1-Mbyte of SDRAM, and entry numbers havebit 16 set to ‘1’. Address of first word of entry in CAM is determined by concatenating entry number with 3 bits of ‘0’. -
CPU 12 creates entries in hash table for DstIP addresses by hashing address and using resulting 16-bit hash index as offset to entry in table. When multiple address' hash to same entry in base table, link is created to free buffer pool entry. If additional addresses hash to same location, they can be added to end of linked list.CPU 12 creates and maintains entries and manages linked list structures. - Control memory block (CTL MEM)136 uses 128K×16 synchronous SRAM (SSRAM), instead of SDRAM devices because most data structures stored require single read and write accesses.
L3 Engine 70 uses 32-KB portion of control memory to store age table. It does single read followed by single write of word in age table. Each 16-bit word contains age table information for 4 CAM entries. Aging information forparticular L3 CAM 126 entry is accessed by using CAM entry number divided by 4 as address into age table. - Forwarding Engine (FE)52 performs lookup requests to
L3 Engine 70 for each IP packet to be processed. Four-deep FIFO buffer is provided to buffer these requests.FE 52 provides RCV Port Number and Block Number for each packet. After lookup is complete,L3 Engine 70 returns RCV Port Number as well as L3 Result and L3 Status word containing various flags and information from matching CAM entry. - Regarding age table support, since
control memory 136 containing age table does not support locked operations, table modifications are done by hardware. Such table modifications address condition of two agents trying to modify same table entry.CPU 12 can initialize entries to invalid state at startup by writing to control memory; but during operation, hardware performs table modifications. - Age table operations are done by
CPU 12 write to age command register. Write to age command register causes Age Table Busy flag in L3 Status register to be set until operation is complete. Aged entries are reported in registers (e.g., AgeResult 1&2). - In FIG. 15A, age table maintenance is illustrated, starting with CPU or live command processing200, then determine if
age command 202 applies. If so,increment time stamp 210 and set age flag; otherwise, readtable entry 204, mask and modifytable entry 206, and writetable entry 208. Further, in FIG. 15B, after age flag set 212, age table is read 216, then determine age out 218. If so, then write result registers 220, set result valid 222, wait forCPU 224, and clear result valid 226; otherwise determine 228 if last entry. Next,clear age flag 232, 230 and read hash table 240. - Time stamp is 2-bit value providing four age time intervals. There are two age time counters, currTime and ageTime. CurrTime is reset to zero and increments when
CPU 12 issues age command. Entries with time stamps equal to this value are newest entries. AgeTime value is always equal to currTime +1 m(i.e., currTime −3, modulo 4). Entries with time stamps equal to ageTime are agednext time CPU 12 issues age command. - CPU adds entry to age table when creating new entry in
L3 CAM 126. Until entry is added to age table, entry does not participate in aging process.CPU 12 writes (e.g., AgeCmd) register with entry number and add or add permanent command, and hardware reads appropriate entry, modifies valid and permanent bits appropriately and writes currTime into time stamp field. - Hardware makes entry live (i.e., accessed) when L3 CAM lookup results in IP hit. Entry number of matching entry is used to access age table, and time stamp field is updated with currTime. Entries which are accessed frequently have more recent time stamp than infrequently used entries, and are not aged out.
-
CPU 12 deletes entry in age table when removing entry fromL3 CAM 126.CPU 12 writes AgeCmd register with entry number and delete command, and hardware reads appropriate entry, clears valid bit, and writes modified entry back to table. - When
CPU 12 age timer expires, CPU writes AgeCmd register to initiate aging process. This sets AgeCmd Busy bit in L3 Status register until entire table is aged. Add and delete commands can be issued, but new age commands have no affect. - When CPU writes ageCmd register, hardware increments ageTime and currTime counters and resets aging address counter to zero. Hardware reads 32K words of age table and checks if any time stamp fields are equal to ageTime. Entries with time stamps equal to ageTime are reported to
CPU 12 as aged out.CPU 12 deletes aged entry from CAM and age table. - To assist CPU in managing linked CAM entries, hardware reports aged entry number and entry number of previous entry in linked list. For aged entries in base hash table, zero value is reported for previous entry. When result is posted to result registers, Age Result Valid bit is set in L3 Status register, and aging process is halted until result is read by
CPU 12. Reading AgeResult register restarts aging process and clears status register bit. - First aged entry number used to access
L3 CAM 126 to retrieve DstIP for entry. DstIP is hashed to locate base hash table entry and CAM entry at address is read. Hardware follows linked list, reading CAM entries until retrieving entry with Link Address equal to original aged entry number. Entry number is reported along with aged entry number in AgeResult registers. -
CPU 12 provides L3 CAM management functions, including initial setup, adding entries, deleting entries and managing linked lists and free buffer pool. Hardware provides automatic search/change capability to assistCPU 12 in locating entries with certain characteristics and optionally changing such entries. - Search operations are initiated by
CPU 12 write to SearchCmd register. Write to SearchCmd register causes Search Busy flag in L3 Status register to set until operation is complete. Matching entries are reported in (e.g., SearchResult) registers. - FIG. 16A shows search operation steps. CPU initiates
search 234, writes commands 236, initialize entry to zero 238, read hash table 240, then determinematch 242. If so, write result toregisters 244, and wait forCPU 246; otherwise, determine if linked 248. If so,clear age flag 250, else, determine iflast entry 254. If so,clear age flag 252; otherwise,clear age flag 256. - Hardware performs automatic and exhaustive search of
L3 CAM 126 when SearchCmd register is written. Starting withentry 0, each entry in base hash table is read and checked against search criteria. If entries have valid link address, then linked entries are read and checked. Minimum 64K CAM entries are read. - During search, SearchCmd can be written with Abort Flag set, and hardware exits search process. Pending SearchResults are read by
CPU 12 before hardware exists and clears Search Busy flag. - For each of 8 words in CAM entry, there is corresponding (e.g., SearchMask and SearchData) registers (16 registers total). Before search command is issued, SearchMask registers are written. ‘0’ in bit position masks bit from consideration in comparison. SearchData registers are written with data values to be matched. Registers containing data to be matched are written.
- Match is indicated when all eight words of CAM entry meet following representative requirement: SearchMaskX & SearchDataX=SearchMaskX & CamDataX.
- To assist
CPU 12 in managing linked CAM entries, hardware reports entry number where match found and entry number of previous entry in linked list. For entries in base hash table matching search criteria, zero value is reported for previous entry. When result is posted to result registers, (e.g., Search Result Valid) bit is set in L3 Status register, and search is halted until result is read byCPU 12. Reading SearchResult2 register restarts search and clears status register bit. - As hardware searches CAM entries and follows linked lists, it stores address of previous entry in register. Entry number is reported with matching entry number in SearchResult registers.
- If change option was selected when (e.g., SearchCmd) value was written, then matching entries found during search are changed by hardware according to values written to change setup registers. When matching entry is found, hardware alters data and writes back to CAM before reporting match result to
CPU 12. - For each of 8 words in CAM entry, there is corresponding (e.g., ChangeMask and ChangeData) registers (16 registers total). Before search command with change option is issued, ChangeMask registers are written. ‘1’ in bit position marks bits to be changed. ChangeData registers are written with desired new data values. Registers containing data to be changed are written.
- For each match, eight words of CAM entry are changed, for example, as follows:
- NewCamDataX=ChangeMaskX & ChangeDataX|˜ChangeMaskX & CamDataX.
- If (e.g., Don't Report) Flag is set when SearchCmd is written, then matching entries are not reported to
CPU 12. Flag should not be set for search only commands. - L3 Engine receives CAM lookup requests from forwarding engine and searches matching entry in
L3 CAM 126. Results of search are returned toFE 52, and additional requests are serviced. - In FIG. 16B, flow chart shows CAM lookup steps. Initially, valid buffer is set258, read
IP header RAM 260, hash destination IP (DstIP)address 262, and read hash table 264, then determine if valid and hit 266. If so, readport MAC RAM 268, updatepacket data 270, modifypacket 272, and writeresult 274. If not, determine valid link 276. If so, followlink 278 and read hash table 264, else writeresult 274. -
L3 Engine 70 buffers up to 4 lookup requests from forwarding engine (FE) 52. When buffer is full, busy signal is sent toFE 52. Buffer is organized as FIFO and contains receiving port number and block number for lookup request. - When valid request exists in buffer, hardware begins lookup process. First buffered request is read, and receive port number for that request is used to access IP header RAM and retrieve packets DstIP address.
- 32 bit DstIP address is hashed to 16-bit value which is used as entry number for base hash table. That entry is read, and words containing DstIP address are compared to packets DstIP address. If these two addresses match, then IP hit bit is set, and results of successful lookup are returned to
FE 52. - Before result is posted to
FE 52, packet may be modified, depending on bit in L3 Flags field of CAM entry. If Don't modify bit of CAM entry is set, nothing is changed in packet. Otherwise, when lookup is successful, TTL field of IP header is decremented and modified in packet memory, and (e.g., CheckSum) field is recalculated and changed. Packets DA is overwritten with value contained in matching CAM entry, and SA is replaced with value from MAC Address Registers and MAC Address RAM. - Whenever IP addresses don't match or CAM entry is not valid, hardware checks Link Valid field of entry to see if entries with same hash index exist. If link valid bit is set, each entry in linked list is read and checked for matching IP address until hit occurs or end of list is reached.
- If match is not found in
L3 CAM 126, hardware checks to see if default route registers are written by CPU. These registers provide ‘default route’ CAM entry and are programmed with same information as CAM entries incontrol memory 136. If default route exists, then packet is modified using default information, and (e.g., IPHit and DefaultRouteUsed) bits of L3 Result are set. - Upper bits of MAC address for ports are provided in three registers. Three 16-bit registers provide full 48-bits address, but lower byte of address for each port is provided by
MAC Address RAM 196. -
MAC Address RAM 196 contains lower byte of MAC address for each port. It is 32×8 dual port ram which is written byCPU 12 and read by hardware during packet modification. This value replaces lower byte from MAC Address Registers when writing new SA for packet. - L3 results of CAM lookup returned to FE include receive port number and block number originally provided by
FE 52 and two 16-bit values, L3 Result and L3 Status. A detailed bit definition for these last two values was provided earlier in this document. -
Switch circuit 20 operates on various performance and integrity levels. In cut-thru switching mode, relatively fast ether switching mode and high performance level are achieved; however, there is possibility of transmitting packet with error during receive process. In this mode, CPU programs MACs to raise receive request when collecting 64 bytes. Also it programs MAC to raise subsequent receive request after every 64 bytes collection. First request provides fast header analysis and switching. - In store forward (SF) mode, receiving packet is not sent to port until reception is complete.
Switch circuit 20 waits until packet completion and updates transmit queues of relevant destination. MAC programming remains same. Forwarding Block acts on ‘store forward mode’ mode on port by port basis. SF mode is selectable as per port basis. In this mode port linking is disabled. - In preferred embodiment, packet moves in following directions: received on LAN port(x) and transmitted to any/all other LAN ports; received on LAN port(x) and posted to UL queue; received on LAN port(x) and posted to CPU queue; received on LAN port(x) and packet dropped; and forwarded from CPU to any/all LAN ports. In each case, packet flows through
packet memory 16 andswitch circuit 20. In each case,switch circuit 20 participates in forwarding, filtering and queue management. - In case of Ethernet port originated packet flow, packet is received on Ethernet ports, and
switch circuit 20 is triggered on such packet from request from one of MACs. This is hardware trigger mode.Switch circuit 20, in coordination withRISC processor 12, allocates free block pulled from receive free list. Once block is assigned, block is busy until destination agent(s) complete transmission. Transmission completion has mechanism to release block and insert in receive free list. - To identify packet destination,
switch circuit 20 needs to obtain packet header information. Header is extracted from MAC received data stream. Here,PM engine 90 identifies header from data stream and loads on port-specific segment of Ether header memory. CAM Processor makes lookup on CAM and delivers result to Auto Forwarding Block. AFB adds packet to one of following queues: one of Ethernet ports transmit queue; all Ethernet ports queues, UL transmit queue and CPU queue; UL queue; CPU queue; or L3 block for L3 lookup. - AFB handles updating “Block address”, “Byte count”, and “routing information” on transmit queues. Once such information is provided, respective transmitting agents handle packet transmission. At end of transmission, block is released and added to receive free list. Refer for more details about releasing the block in some other sections.
-
CPU 12 posts packets toXQMB 154 of Auto Forwarding Block for transmission to one or several ports.XQMB 154 handles posting packet to respective queues. Prior to request,CPU 12 assembles or modifies existing packet inpacket memory 16 for transmission.CPU 12, with help of DMA function, can transfer packet fromlocal memory 100 topacket memory 16, and at end of such transfer, can initiateXQMB 154 action. - FIG. 17 shows packet receive process, which is accomplished by Receive
Arbiter 80,Buffer scheduler 94, LAN bus controller 76,Packet memory engine 90,header memory 290,data buffer 84, as well as Auto Forwarding Block (AFB) and PM SDRAM controller. Such modules work concurrently. - Receive
arbiter 80 arbitrates and prioritizes receive request from Ether ports. It raises request to bufferscheduler 94. When request is under process,arbiter 80 makes background processing on remaining requests. -
Buffer scheduler 94 handles resource allocation of internal buffers.Buffer scheduler 94 maintains two receive buffers and two transmitbuffers 84. Each buffer can hold up to 64 bytes of data, and optimizes buffer allocation algorithm for fair bandwidth extraction between receivers/transmitters. - LAN Bus Controller76 interfaces to LAN bus to read/write packet data to MAC FIFO buffers, and access MAC receive completion status. LAN bus controller 76 may access MAC and read data slice, and store to internal data buffer. A data buffer can hold up to 64 bytes of information.
-
Packet memory engine 90 sets-up moving packet in slice-by-slice intoPM 16.Packet memory engine 90 reads PM receive block register and byte count register, and updates (i.e., increments) byte count register on each transfer.Packet memory engine 90 commands PM SDRAM controller to start data transfer. PM SDRAM controller transfers data from receive buffer topacket memory 16, and generates control timing to access external SDRAM. - Auto Forwarding Block allocates free block to receiver; initializes receive block register, byte count register. Occasionally AFB commands to reject packet.
- After reset, if (e.g., AUTO_IMT_STRT) sense bit is cleared, (e.g., RCVR_INIT) receiver initialize block in
FE 52 waits for (e.g., INIT_STRT) initialize start bit to be set. When bit is set, then using (e.g., PORT_ES_STS) information, which tells receive ports to be initialized, active receive ports are initialized with free block. - Every LAN MAC has (e.g., RREQx) receive request signal which, when active, including at least 64 bytes of data (i.e., header/data region) is collected in internal FIFO. There are 32 request signals from LAN bus. Following steps describe new packet reception and header memory loading:
- 1. When new packet starts in Port-x, RREQx signal becomes active, indicating 64 bytes valid in FIFO. Request active and (e.g., RBODYx) bit clear means new packet.
- 2. RREQx signal is first-level conditioned if corresponding bit enabled in Receive Enable register. When
RISC 12 allocates free block to receive port, it writes block address on corresponding receive block register and enables the receiver”. Conditional RREQx first wins RREQ arbitration to get service. - 3. When RREQx wins arbitration, scanner freezes on port number x and request service to buffer
scheduler 94. - 4.
Buffer scheduler 94 allocates one of two free receive data buffer and enables LAN bus controller 76 to start data transfer. - 5. LAN bus controller76 executes Burst Read Accesses on LAN bus targeted to Port-x. Read data is written on allocated internal receive buffer. Since body bit is clear, loading process signatures slice as “header”. If slice is header, it writes header data on 2-port Ether Header memory. At end of header loading, Port to be Analyzed FIFO is loaded with 5-bit port number. Loading of FIFO enables CAM engine to start analyzing header information. Load completion calls attention of PM engine for data movement from receive data buffer to
packet memory 16. - 6.
PM engine 90 updates byte count and sets-up SDRAM Controller for data transfer topacket memory 16. - 7. Buffers have dedicated channels to SDRAM Controller. SDRAM Controller arbitrates transfer requests amongst channels and starts executing request at time overlapping address and data phases to maximize throughput and efficiency. Requesting channel is held arbitrating for LAN bus until full slice is moved in
packet memory 16. - Packet data reception procedure, similarly to MAC RREQx signal which, when active informs at least 64 bytes data, is collected in internal FIFO, as follows:
- 1. When body bit for Port(x) is set, and RREQx signal is active, 64 bytes are valid in FIFO.
- 2. RREQx signal is first-level conditioned if corresponding bit is enabled in Receive Enable register. Conditional RREQx first wins the RREQ SCAN arbitration to get the service.
- 3. When RREQx wins arbitration, scanner freezes on port number, (x) and requests service to buffer
scheduler 94. Along with this request, RBODYx bit indicates if packet is in middle of reception. - 4.
Buffer scheduler 94 allocates free receivedata buffer 84 and alerts LAN bus controller 76. - 5. After acquiring LAN bus, LAN bus controller76 executes burst-read accesses on LAN bus targeted to Port-x. Read data is written in allocated receive
buffer 84. - 6. When data loading is completed on data buffer, loaded data buffer is signatured as “non header”. LAN bus controller76 continues scan for next request in queue. Loaded data buffer draws attention of
PM engine 90 to load toPM 16. - When processing in unswitched environment, (e.g., Rec Link) receiver link register(x) indicates accordingly. Link bit cleared means slice is non-switch mode presently. Loaded data buffer calls attention of
PM engine 90 to load data intopacket memory 16.PM engine 90 uses the Receiver Block Address register(x) and Receiver BC register(x) to construct PM destination address. It leaves updated byte count in Rec BC register(x) for future reference. PM engine issues command to PM SDRAM controller to start data transfer and ready to service receive or transmit buffer or accept command. - When processing in switched environment, Rec Link register(x) indicates accordingly. If link bit is set, slice is in switch mode for present slice and consecutive slices until end of packet. Loaded data buffer calls attention of
PM engine 90 to load data intopacket memory 16.PM engine 90 uses Receive Block Address register(x) and Receive BC register(x) to construct PM destination address. Byte count is updated in receive BC register(x) and transmit BC register(y). PM engine issues command to PM SDRAM controller to start data transfer and ready to service receive or transmit buffer or accept command. - Forwarding Engine (FE), in coordination with CAM Processor and L3 Lookup Block, evaluates current receiving packet for following possible decisions: reject packet; link packet; forward packet to transmitter queue; multicast packet to two or more ports; broadcast packet only to Ether ports; broadcast packet to Ether ports and UL; send packet to UL; send packet to CPU; or send packet to L3 Lookup Block for L3 analysis.
- In packet forwarding mode, CAM Processor writes decision information into header analyzed FIFO. Such write process wakes-up Forwarding Block to take-up forwarding process.
PM engine 90 keeps loading successive slices of packet inpacket memory 16 independent of CAM analysis. Decision of CAM might occur at middle of packet reception or after end of packet reception. If header analysis is complete before packet reception is complete, Forwarding Block acts on packet if packet is unicast or destination is L3 Lookup Block which carries on further analysis associated with L3 forwarding. For other cases, Forwarding Block is not called to action until receive completion of packet. If packet reception is complete before header analysis, Forwarding Block is not called into action until header analysis is complete. Receiver is not primed again until forwarding decision has been taken on received block and acted upon. - In packet rejection mode, when CAM engine or Forwarding Block decides not to receive packet, control bit is set for corresponding receiving port to reject incoming packet.
PM engine 90 looks at reject bit to set prepare for transfer from receive buffers to packet memory. If reject is set,PM engine 90 empties FIFO without setting transfer to PM.PM engine 90 clears reject bit at end of packet reception. Receive complete state is indicated to Forwarding Block. - In packet switching mode, CAM lookup posts port number in CAM Analysis Done FIFO in addition to setting CAM analysis done bit for port. This draws attention of Forwarding Block prior to completion of packet reception. Forwarding Block checks several conditions to take forwarding action. At this time, it may link packet to corresponding transmitter or post packet in queue of transmitter.
- Transmitter may be busy, i.e., transmitter queue contains one or more packets queued or transmitter is currently processing old packet. Forwarding Block requests
XQMB 154 to post receive packet in transmitter queue with incomplete information. This is handled by manipulating RC bit clear in BC entry incontrol memory 136. Bit, if clear, means packet block address is valid, but byte count is invalid. Packet data is incomplete inpacket memory 16. Forwarding Block pushes incomplete packet in transmitter queue on special occasion. When, receiver(x) wants to switch to transmitter and if transmitter is currently busy, Forwarding Block puts packet in transmitter queue to maintain order of priority. At pushing event, byte count information is invalid. If transmitter finishes old packet, and receiver/packet falls as next packet in transmitter queue, and receiver has not completed full reception, thenXQMB 154 commands to link. If receiver completes packet before getting to transmitter, Forwarding Block sets such bit, and loads valid BC value on BC entry. Subsequent forwarding action on packet degenerates to store and forward mode. - Transmitter may be free when switching decision occurs. Forwarding Block commands to link receiver to transmitter. It does not manipulate control memory structure. At receive packet completion, Forwarding Block primes receiver, and transmitter continues to transmit until end of packet without further intervention. Finishing transmitter event releases block and pushes to receive free list.
- Forwarding Block may act on receive packet after receive completion and CAM analysis completion. Since packet is received,
XQMB 154 is instructed to post on appropriate transmitter queue; this is Store and Forward mode. - Broadcasting to Ether ports decision may result from not finding destination port or hit on broadcast MAC address. Depending on nature of broadcast decision, broadcast map for receiving port is fetched, and packet is forwarded to transmit queue management block for posting on transmit queues
- In sending packet to uplink or
processor 12, header analysis results in port not physically connected to LAN ports, but CPU or uplink ports; Forwarding Engine Block instructsXQMB 154 to queue on appropriate ports.XQMB 154 may queue packet on ports if ports are specified in broadcast port map. - When end of packet is sensed from LAN port(x), LAN controller signals by bit in slice status.
PM engine 90, while moving slice toPM 16, notifies same status by setting appropriate bit in Rec end reg. Forwarding Block acts on every receive completion; and in addition to forwarding actions, it instructs Free Queue Management Block to prime receiver. In case previous packet is rejected, no new block needs to be allocated; in such case it enables receiver to receive new packet. - For packet transmission, transmit port activity is top-level enabled by Forwarding Block or Transmit Queue Management Block of Auto Forwarding Block.
- In stored packet mode from transmit queue,
XQMB 154 picks highest priority packet, and loads transmit block address register and byte count register corresponding to packet. This action enables transmitter on transmit enable register. - In cut-thru mode for linking, Forwarding Block or
XQMB 154 loads link command with which hardware copies receiver block address to transmitter block address register. It copies current running receiver rec(x) byte count value to transmitter(y) byte count register. It also sets link bit active. - Transmitter enters arbitration if transmit (e.g., XMT) enable bit is set, and Byte count validity is met. If MAC transmit FIFO has at least 64 bytes free space, it raises TREQ# signal. This signal is conditioned with first phase enable signal, and transmitter enters arbitration with other TREQ# signals asserted by other transmitters. Winning transmitter requests allocation of one of two free transmit buffers. This request is forwarded to buffer scheduler. When buffer scheduler allocates free buffer, transmitter enters arbitration for
PM engine 90 service.PM engine 90 time-multiplexes between receive requests and transmit requests and other commands such as link, receive enable and transmit enable. -
PM engine 90 moves sets-up transfer with PM SDRAM controller by giving command to move slice from PM to data buffer, and updates byte count and address registers in array for corresponding transmitter. -
PM engine 90 signatures slice as header or non-header based on XMT body bit. Along with slice,PM engine 90 passes information, such as slice count and port address through buffer attributes. Loaded slice calls attention of corresponding LAN controller for service to transfer data from transmit buffer to MAC on LAN bus. LAN controller moves slice to target MAC port and releases buffer. WheneverPM engine 90 moves slice, decremented byte count is compared, to check, if reached zero. If reached zero, packet may reach end of packet status based on following cases: - If case is non-linked, packet was originally fully received. Byte count loaded at time of enable was actual byte count of packet. Link bit clear indicates end of packet.
- If packet is in linked state, transmitter byte count reaching zero is not regarded as end of packet; it is regarded as transmitter has to wait for receiver to get slice. Transmitter does not participate in arbitration again until slice is received on linked receiver. Link bit clear and byte count zero signal packet completion.
- PM Engine clears XMT enable bit, and sets End of packet transmit bit. End of transmit draws attention of
XQMB 154 to look at transmitter queue incontrol memory 136. If queue contains additional packets,XQMB 154 loads new packet to re-enable transmitter. If queue is empty,XQMB 154 does not take action. Trigger point for enabling transmitter is: when current packet ends, and new packet is pending in queue; when receive packet is targeted to transmitter and queue is empty; or when CPU inserts packet to transmitter. - Foregoing described embodiments of invention are provided as illustration and description. It is not intended to limit invention to precise form described. Such described specification contemplates that inventive functionality may be equivalently implemented in software, firmware, hardware, and/or other functionally comparable or equivalent electronic digital processing system or circuit made available to one of ordinary skill in the art. Other variations and embodiments are possible in light of above teaching, and it is thus intended that scope of invention not be limited by detailed description, but rather by claims as follow.
Claims (22)
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US09/118,458 US6424659B2 (en) | 1998-07-17 | 1998-07-17 | Multi-layer switching apparatus and method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
US09/118,458 US6424659B2 (en) | 1998-07-17 | 1998-07-17 | Multi-layer switching apparatus and method |
Publications (2)
Publication Number | Publication Date |
---|---|
US20010043614A1 true US20010043614A1 (en) | 2001-11-22 |
US6424659B2 US6424659B2 (en) | 2002-07-23 |
Family
ID=22378721
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
US09/118,458 Expired - Lifetime US6424659B2 (en) | 1998-07-17 | 1998-07-17 | Multi-layer switching apparatus and method |
Country Status (1)
Country | Link |
---|---|
US (1) | US6424659B2 (en) |
Cited By (157)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20020105958A1 (en) * | 2000-12-12 | 2002-08-08 | Jean-Pierre Mao | Process and device for deterministic transmission of asynchronous data in packets |
US20020120888A1 (en) * | 2001-02-14 | 2002-08-29 | Jorg Franke | Network co-processor for vehicles |
US20020143910A1 (en) * | 2001-03-29 | 2002-10-03 | Shih-Wei Chou | Network hub |
US20030007488A1 (en) * | 2001-06-22 | 2003-01-09 | Sriram Rao | Efficient data transmission based on a policy |
US20030041216A1 (en) * | 2001-08-27 | 2003-02-27 | Rosenbluth Mark B. | Mechanism for providing early coherency detection to enable high performance memory updates in a latency sensitive multithreaded environment |
US20030095567A1 (en) * | 2001-11-20 | 2003-05-22 | Lo Man Kuk | Real time protocol packet handler |
US20030210688A1 (en) * | 2002-05-13 | 2003-11-13 | International Business Machines Corporation | Logically grouping physical ports into logical interfaces to expand bandwidth |
US20030225965A1 (en) * | 2002-06-04 | 2003-12-04 | Ram Krishnan | Hitless restart of access control module |
US6674769B1 (en) * | 2000-03-07 | 2004-01-06 | Advanced Micro Devices, Inc. | Simultaneous searching of layer 3 policy filter and policy cache in a network switch port |
US6757742B1 (en) * | 2000-05-25 | 2004-06-29 | Advanced Micro Devices, Inc. | Computer-based system for validating hash-based table lookup schemes in a network switch |
US20040139290A1 (en) * | 2003-01-10 | 2004-07-15 | Gilbert Wolrich | Memory interleaving |
US6778526B1 (en) * | 2000-12-22 | 2004-08-17 | Nortel Networks Limited | High speed access bus interface and protocol |
US20040172346A1 (en) * | 2002-08-10 | 2004-09-02 | Cisco Technology, Inc., A California Corporation | Generating accounting data based on access control list entries |
US6791983B1 (en) * | 2000-06-14 | 2004-09-14 | Mindspeed Technologies, Inc. | Content-addressable memory for use with a communication packet processor to retrieve context information |
US6798778B1 (en) * | 2000-06-14 | 2004-09-28 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for updating context information for a core processor |
US6826180B1 (en) * | 2000-06-14 | 2004-11-30 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for storing summation blocks of context information for a core processor |
US6845099B2 (en) * | 2000-06-14 | 2005-01-18 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for modifying selectors to retrieve context information for a core processor |
US6859459B1 (en) * | 1999-11-16 | 2005-02-22 | Nec Corporation | High-speed/high-reliability ether transmission system and I/F apparatus |
US6885666B1 (en) * | 2000-08-14 | 2005-04-26 | Advanced Micro Devices, Inc. | Apparatus and method in a network switch for synchronizing transfer of a control tag to a switch fabric with transfer of frame data to a buffer memory |
US6891829B1 (en) * | 2000-06-14 | 2005-05-10 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for retrieving context information for a core processor |
US20050141501A1 (en) * | 1999-03-17 | 2005-06-30 | Broadcom Corporation | Network switch having a programmable counter |
US6925641B1 (en) * | 2000-02-04 | 2005-08-02 | Xronix Communications, Inc. | Real time DSP load management system |
US6934260B1 (en) * | 2000-02-01 | 2005-08-23 | Advanced Micro Devices, Inc. | Arrangement for controlling learning of layer 3 network addresses in a network switch |
US6940854B1 (en) * | 2001-03-23 | 2005-09-06 | Advanced Micro Devices, Inc. | Systems and methods for determining priority based on previous priority determinations |
US20050198429A1 (en) * | 2004-03-02 | 2005-09-08 | Nec Electronics Corporation | Multilayer system and clock control method |
US20060114895A1 (en) * | 2004-11-30 | 2006-06-01 | Broadcom Corporation | CPU transmission of unmodified packets |
US20060120366A1 (en) * | 2000-09-12 | 2006-06-08 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US20060256787A1 (en) * | 2000-06-23 | 2006-11-16 | Broadcom Corporation | Switch having external address resolution interface |
US7215637B1 (en) * | 2000-04-17 | 2007-05-08 | Juniper Networks, Inc. | Systems and methods for processing packets |
US20070118677A1 (en) * | 2005-05-13 | 2007-05-24 | Freescale Semiconductor Incorporated | Packet switch having a crossbar switch that connects multiport receiving and transmitting elements |
US20070136331A1 (en) * | 2005-11-28 | 2007-06-14 | Nec Laboratories America | Storage-efficient and collision-free hash-based packet processing architecture and method |
US20080126572A1 (en) * | 2006-10-05 | 2008-05-29 | Holt John M | Multi-path switching networks |
US20080144785A1 (en) * | 2006-12-19 | 2008-06-19 | Dae-Hyun Lee | Call setup method and terminal in a IP network |
US20080225837A1 (en) * | 2007-03-16 | 2008-09-18 | Novell, Inc. | System and Method for Multi-Layer Distributed Switching |
US20090037629A1 (en) * | 2007-08-01 | 2009-02-05 | Broadcom Corporation | Master slave core architecture with direct buses |
US7688324B1 (en) * | 1999-03-05 | 2010-03-30 | Zoran Corporation | Interactive set-top box having a unified memory architecture |
US7688727B1 (en) | 2000-04-17 | 2010-03-30 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US7720055B2 (en) * | 1999-03-17 | 2010-05-18 | Broadcom Corporation | Method for handling IP multicast packets in network switch |
US20100153961A1 (en) * | 2004-02-10 | 2010-06-17 | Hitachi, Ltd. | Storage system having processor and interface adapters that can be increased or decreased based on required performance |
US20110080913A1 (en) * | 2006-07-21 | 2011-04-07 | Cortina Systems, Inc. | Apparatus and method for layer-2 to 7 search engine for high speed network application |
US20110110372A1 (en) * | 2003-06-05 | 2011-05-12 | Juniper Networks, Inc. | Systems and methods to perform hybrid switching and routing functions |
WO2013026049A1 (en) * | 2011-08-17 | 2013-02-21 | Nicira, Inc. | Distributed logical l3 routing |
US20130182707A1 (en) * | 2012-01-18 | 2013-07-18 | International Business Machines Corporation | Managing a global forwarding table in a distributed switch |
US20140032701A1 (en) * | 2009-02-19 | 2014-01-30 | Micron Technology, Inc. | Memory network methods, apparatus, and systems |
US8717895B2 (en) | 2010-07-06 | 2014-05-06 | Nicira, Inc. | Network virtualization apparatus and method with a table mapping engine |
US20140140346A1 (en) * | 2012-11-22 | 2014-05-22 | Hitachi Metals, Ltd. | Communication System and Network Relay Device |
US8856419B2 (en) | 2010-07-19 | 2014-10-07 | International Business Machines Corporation | Register access in distributed virtual bridge environment |
US8861400B2 (en) | 2012-01-18 | 2014-10-14 | International Business Machines Corporation | Requesting multicast membership information in a distributed switch in response to a miss event |
US20140351885A1 (en) * | 2013-05-22 | 2014-11-27 | Unisys Corporation | Control of simple network management protocol activity |
US8966035B2 (en) | 2009-04-01 | 2015-02-24 | Nicira, Inc. | Method and apparatus for implementing and managing distributed virtual switches in several hosts and physical forwarding elements |
US8964528B2 (en) | 2010-07-06 | 2015-02-24 | Nicira, Inc. | Method and apparatus for robust packet distribution among hierarchical managed switching elements |
US9043452B2 (en) | 2011-05-04 | 2015-05-26 | Nicira, Inc. | Network control apparatus and method for port isolation |
US9083609B2 (en) | 2007-09-26 | 2015-07-14 | Nicira, Inc. | Network operating system for managing and securing networks |
US9225597B2 (en) | 2014-03-14 | 2015-12-29 | Nicira, Inc. | Managed gateways peering with external router to attract ingress packets |
US9306864B2 (en) | 2011-10-25 | 2016-04-05 | Nicira, Inc. | Scheduling distribution of physical control plane data |
US9306910B2 (en) | 2009-07-27 | 2016-04-05 | Vmware, Inc. | Private allocated networks over shared communications infrastructure |
US9313129B2 (en) | 2014-03-14 | 2016-04-12 | Nicira, Inc. | Logical router processing by network controller |
US9385954B2 (en) | 2014-03-31 | 2016-07-05 | Nicira, Inc. | Hashing techniques for use in a network environment |
US9407580B2 (en) | 2013-07-12 | 2016-08-02 | Nicira, Inc. | Maintaining data stored with a packet |
US9413644B2 (en) | 2014-03-27 | 2016-08-09 | Nicira, Inc. | Ingress ECMP in virtual distributed routing environment |
US9419855B2 (en) | 2014-03-14 | 2016-08-16 | Nicira, Inc. | Static routes for logical routers |
US9444651B2 (en) | 2011-08-17 | 2016-09-13 | Nicira, Inc. | Flow generation from second level controller to first level controller to managed switching element |
US9503371B2 (en) | 2013-09-04 | 2016-11-22 | Nicira, Inc. | High availability L3 gateways for logical networks |
US9503321B2 (en) | 2014-03-21 | 2016-11-22 | Nicira, Inc. | Dynamic routing for logical routers |
US9525647B2 (en) | 2010-07-06 | 2016-12-20 | Nicira, Inc. | Network control apparatus and method for creating and modifying logical switching elements |
US9548924B2 (en) | 2013-12-09 | 2017-01-17 | Nicira, Inc. | Detecting an elephant flow based on the size of a packet |
US9559987B1 (en) * | 2008-09-26 | 2017-01-31 | Tellabs Operations, Inc | Method and apparatus for improving CAM learn throughput using a cache |
US9569368B2 (en) | 2013-12-13 | 2017-02-14 | Nicira, Inc. | Installing and managing flows in a flow table cache |
US9571386B2 (en) | 2013-07-08 | 2017-02-14 | Nicira, Inc. | Hybrid packet processing |
US9575782B2 (en) | 2013-10-13 | 2017-02-21 | Nicira, Inc. | ARP for logical router |
US9577845B2 (en) | 2013-09-04 | 2017-02-21 | Nicira, Inc. | Multiple active L3 gateways for logical networks |
US9590901B2 (en) | 2014-03-14 | 2017-03-07 | Nicira, Inc. | Route advertisement by managed gateways |
US20170070474A1 (en) * | 2013-10-06 | 2017-03-09 | Mellanox Technologies Ltd. | Simplified packet routing |
US9602398B2 (en) | 2013-09-15 | 2017-03-21 | Nicira, Inc. | Dynamically generating flows with wildcard fields |
US9647883B2 (en) | 2014-03-21 | 2017-05-09 | Nicria, Inc. | Multiple levels of logical routers |
US9680750B2 (en) | 2010-07-06 | 2017-06-13 | Nicira, Inc. | Use of tunnels to hide network addresses |
US9697032B2 (en) | 2009-07-27 | 2017-07-04 | Vmware, Inc. | Automated network configuration of virtual machines in a virtual lab environment |
US9742881B2 (en) | 2014-06-30 | 2017-08-22 | Nicira, Inc. | Network virtualization using just-in-time distributed capability for classification encoding |
US9768980B2 (en) | 2014-09-30 | 2017-09-19 | Nicira, Inc. | Virtual distributed bridging |
US9887960B2 (en) | 2013-08-14 | 2018-02-06 | Nicira, Inc. | Providing services for logical networks |
US9893988B2 (en) | 2014-03-27 | 2018-02-13 | Nicira, Inc. | Address resolution using multiple designated instances of a logical router |
US9900410B2 (en) | 2006-05-01 | 2018-02-20 | Nicira, Inc. | Private ethernet overlay networks over a shared ethernet in a virtual environment |
US9923760B2 (en) | 2015-04-06 | 2018-03-20 | Nicira, Inc. | Reduction of churn in a network control system |
AU2014398480B2 (en) * | 2014-06-24 | 2018-04-05 | Hitachi, Ltd. | Financial products trading system and financial products trading control method |
US9952885B2 (en) | 2013-08-14 | 2018-04-24 | Nicira, Inc. | Generation of configuration files for a DHCP module executing within a virtualized container |
US9967199B2 (en) | 2013-12-09 | 2018-05-08 | Nicira, Inc. | Inspecting operations of a machine to detect elephant flows |
US9996467B2 (en) | 2013-12-13 | 2018-06-12 | Nicira, Inc. | Dynamically adjusting the number of flows allowed in a flow table cache |
US10020960B2 (en) | 2014-09-30 | 2018-07-10 | Nicira, Inc. | Virtual distributed bridging |
US10033579B2 (en) | 2012-04-18 | 2018-07-24 | Nicira, Inc. | Using transactions to compute and propagate network forwarding state |
US10038628B2 (en) | 2015-04-04 | 2018-07-31 | Nicira, Inc. | Route server mode for dynamic routing between logical and physical networks |
US10057157B2 (en) | 2015-08-31 | 2018-08-21 | Nicira, Inc. | Automatically advertising NAT routes between logical routers |
US10063458B2 (en) | 2013-10-13 | 2018-08-28 | Nicira, Inc. | Asymmetric connection with external networks |
US10079779B2 (en) | 2015-01-30 | 2018-09-18 | Nicira, Inc. | Implementing logical router uplinks |
US10091161B2 (en) | 2016-04-30 | 2018-10-02 | Nicira, Inc. | Assignment of router ID for logical routers |
US10089127B2 (en) | 2011-11-15 | 2018-10-02 | Nicira, Inc. | Control plane interface for logical middlebox services |
US10095535B2 (en) | 2015-10-31 | 2018-10-09 | Nicira, Inc. | Static route types for logical routers |
US10103939B2 (en) | 2010-07-06 | 2018-10-16 | Nicira, Inc. | Network control apparatus and method for populating logical datapath sets |
US10129142B2 (en) | 2015-08-11 | 2018-11-13 | Nicira, Inc. | Route configuration for logical router |
US10153973B2 (en) | 2016-06-29 | 2018-12-11 | Nicira, Inc. | Installation of routing tables for logical router in route server mode |
US20190007364A1 (en) * | 2017-06-30 | 2019-01-03 | Nicira, Inc. | Assignment of unique physical network addresses for logical network addresses |
US10178029B2 (en) | 2016-05-11 | 2019-01-08 | Mellanox Technologies Tlv Ltd. | Forwarding of adaptive routing notifications |
US10181993B2 (en) | 2013-07-12 | 2019-01-15 | Nicira, Inc. | Tracing network packets through logical and physical networks |
US10193806B2 (en) | 2014-03-31 | 2019-01-29 | Nicira, Inc. | Performing a finishing operation to improve the quality of a resulting hash |
US10200306B2 (en) | 2017-03-07 | 2019-02-05 | Nicira, Inc. | Visualization of packet tracing operation results |
US10200294B2 (en) | 2016-12-22 | 2019-02-05 | Mellanox Technologies Tlv Ltd. | Adaptive routing based on flow-control credits |
US10204122B2 (en) | 2015-09-30 | 2019-02-12 | Nicira, Inc. | Implementing an interface between tuple and message-driven control entities |
US10212071B2 (en) | 2016-12-21 | 2019-02-19 | Nicira, Inc. | Bypassing a load balancer in a return path of network traffic |
US20190065494A1 (en) * | 2017-08-28 | 2019-02-28 | International Business Machines Corporation | Efficient and accurate lookups of data by a stream processor using a hash table |
US10225184B2 (en) | 2015-06-30 | 2019-03-05 | Nicira, Inc. | Redirecting traffic in a virtual distributed router environment |
US10237123B2 (en) | 2016-12-21 | 2019-03-19 | Nicira, Inc. | Dynamic recovery from a split-brain failure in edge nodes |
US10250443B2 (en) | 2014-09-30 | 2019-04-02 | Nicira, Inc. | Using physical location to modify behavior of a distributed virtual network element |
US10333849B2 (en) | 2016-04-28 | 2019-06-25 | Nicira, Inc. | Automatic configuration of logical routers on edge nodes |
US10341236B2 (en) | 2016-09-30 | 2019-07-02 | Nicira, Inc. | Anycast edge service gateways |
US10374827B2 (en) | 2017-11-14 | 2019-08-06 | Nicira, Inc. | Identifier that maps to different networks at different datacenters |
US10454758B2 (en) | 2016-08-31 | 2019-10-22 | Nicira, Inc. | Edge node cluster network redundancy and fast convergence using an underlay anycast VTEP IP |
US10469342B2 (en) | 2014-10-10 | 2019-11-05 | Nicira, Inc. | Logical network traffic analysis |
US10484515B2 (en) | 2016-04-29 | 2019-11-19 | Nicira, Inc. | Implementing logical metadata proxy servers in logical networks |
US10498638B2 (en) | 2013-09-15 | 2019-12-03 | Nicira, Inc. | Performing a multi-stage lookup to classify packets |
US10511459B2 (en) | 2017-11-14 | 2019-12-17 | Nicira, Inc. | Selection of managed forwarding element for bridge spanning multiple datacenters |
US10511458B2 (en) | 2014-09-30 | 2019-12-17 | Nicira, Inc. | Virtual distributed bridging |
US10560320B2 (en) | 2016-06-29 | 2020-02-11 | Nicira, Inc. | Ranking of gateways in cluster |
US10608887B2 (en) | 2017-10-06 | 2020-03-31 | Nicira, Inc. | Using packet tracing tool to automatically execute packet capture operations |
US10616045B2 (en) | 2016-12-22 | 2020-04-07 | Nicira, Inc. | Migration of centralized routing components of logical router |
US10637800B2 (en) | 2017-06-30 | 2020-04-28 | Nicira, Inc | Replacement of logical network addresses with physical network addresses |
US10644995B2 (en) | 2018-02-14 | 2020-05-05 | Mellanox Technologies Tlv Ltd. | Adaptive routing in a box |
US10659373B2 (en) | 2014-03-31 | 2020-05-19 | Nicira, Inc | Processing packets according to hierarchy of flow entry storages |
US10742746B2 (en) | 2016-12-21 | 2020-08-11 | Nicira, Inc. | Bypassing a load balancer in a return path of network traffic |
US10797998B2 (en) | 2018-12-05 | 2020-10-06 | Vmware, Inc. | Route server for distributed routers using hierarchical routing protocol |
US10819621B2 (en) | 2016-02-23 | 2020-10-27 | Mellanox Technologies Tlv Ltd. | Unicast forwarding of adaptive-routing notifications |
US10841273B2 (en) | 2016-04-29 | 2020-11-17 | Nicira, Inc. | Implementing logical DHCP servers in logical networks |
US10931560B2 (en) | 2018-11-23 | 2021-02-23 | Vmware, Inc. | Using route type to determine routing protocol behavior |
US10938788B2 (en) | 2018-12-12 | 2021-03-02 | Vmware, Inc. | Static routes for policy-based VPN |
US11005724B1 (en) | 2019-01-06 | 2021-05-11 | Mellanox Technologies, Ltd. | Network topology having minimal number of long connections among groups of network elements |
US11019167B2 (en) | 2016-04-29 | 2021-05-25 | Nicira, Inc. | Management of update queues for network controller |
US11095480B2 (en) | 2019-08-30 | 2021-08-17 | Vmware, Inc. | Traffic optimization using distributed edge services |
US11178051B2 (en) | 2014-09-30 | 2021-11-16 | Vmware, Inc. | Packet key parser for flow-based forwarding elements |
US11190463B2 (en) | 2008-05-23 | 2021-11-30 | Vmware, Inc. | Distributed virtual switch for virtualized computer systems |
US11196628B1 (en) | 2020-07-29 | 2021-12-07 | Vmware, Inc. | Monitoring container clusters |
US11201808B2 (en) | 2013-07-12 | 2021-12-14 | Nicira, Inc. | Tracing logical network packets through physical network |
US11336533B1 (en) | 2021-01-08 | 2022-05-17 | Vmware, Inc. | Network visualization of correlations between logical elements and associated physical elements |
US11411911B2 (en) | 2020-10-26 | 2022-08-09 | Mellanox Technologies, Ltd. | Routing across multiple subnetworks using address mapping |
US11451413B2 (en) | 2020-07-28 | 2022-09-20 | Vmware, Inc. | Method for advertising availability of distributed gateway service and machines at host computer |
US11558426B2 (en) | 2020-07-29 | 2023-01-17 | Vmware, Inc. | Connection tracking for container cluster |
US11570090B2 (en) | 2020-07-29 | 2023-01-31 | Vmware, Inc. | Flow tracing operation in container cluster |
US11575594B2 (en) | 2020-09-10 | 2023-02-07 | Mellanox Technologies, Ltd. | Deadlock-free rerouting for resolving local link failures using detour paths |
US11606294B2 (en) | 2020-07-16 | 2023-03-14 | Vmware, Inc. | Host computer configured to facilitate distributed SNAT service |
US11611613B2 (en) | 2020-07-24 | 2023-03-21 | Vmware, Inc. | Policy-based forwarding to a load balancer of a load balancing cluster |
US11616755B2 (en) | 2020-07-16 | 2023-03-28 | Vmware, Inc. | Facilitating distributed SNAT service |
US11677645B2 (en) | 2021-09-17 | 2023-06-13 | Vmware, Inc. | Traffic monitoring |
US11687210B2 (en) | 2021-07-05 | 2023-06-27 | Vmware, Inc. | Criteria-based expansion of group nodes in a network topology visualization |
US11711278B2 (en) | 2021-07-24 | 2023-07-25 | Vmware, Inc. | Visualization of flow trace operation across multiple sites |
US11736436B2 (en) | 2020-12-31 | 2023-08-22 | Vmware, Inc. | Identifying routes with indirect addressing in a datacenter |
US11765103B2 (en) | 2021-12-01 | 2023-09-19 | Mellanox Technologies, Ltd. | Large-scale network with high port utilization |
US11855898B1 (en) * | 2018-03-14 | 2023-12-26 | F5, Inc. | Methods for traffic dependent direct memory access optimization and devices thereof |
US11870682B2 (en) | 2021-06-22 | 2024-01-09 | Mellanox Technologies, Ltd. | Deadlock-free local rerouting for handling multiple local link failures in hierarchical network topologies |
US11902050B2 (en) | 2020-07-28 | 2024-02-13 | VMware LLC | Method for providing distributed gateway service at host computer |
US11924080B2 (en) | 2020-01-17 | 2024-03-05 | VMware LLC | Practical overlay network latency measurement in datacenter |
Families Citing this family (123)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6463470B1 (en) | 1998-10-26 | 2002-10-08 | Cisco Technology, Inc. | Method and apparatus of storing policies for policy-based management of quality of service treatments of network data traffic flows |
US7194554B1 (en) | 1998-12-08 | 2007-03-20 | Nomadix, Inc. | Systems and methods for providing dynamic network authorization authentication and accounting |
US8713641B1 (en) | 1998-12-08 | 2014-04-29 | Nomadix, Inc. | Systems and methods for authorizing, authenticating and accounting users having transparent computer access to a network using a gateway device |
US8266266B2 (en) | 1998-12-08 | 2012-09-11 | Nomadix, Inc. | Systems and methods for providing dynamic network authorization, authentication and accounting |
US7382736B2 (en) | 1999-01-12 | 2008-06-03 | Mcdata Corporation | Method for scoring queued frames for selective transmission through a switch |
US6639915B1 (en) * | 1999-04-07 | 2003-10-28 | Utstarcom, Inc. | Method and apparatus for transmission of voice data in a network structure |
US6587461B1 (en) * | 1999-06-08 | 2003-07-01 | Cisco Technology, Inc. | TDM switching system and ASIC device |
US6434703B1 (en) | 1999-06-08 | 2002-08-13 | Cisco Technology, Inc. | Event initiation bus and associated fault protection for a telecommunications device |
US7346677B1 (en) | 1999-07-02 | 2008-03-18 | Cisco Technology, Inc. | Method and apparatus for creating policies for policy-based management of quality of service treatments of network data traffic flows |
US6983350B1 (en) | 1999-08-31 | 2006-01-03 | Intel Corporation | SDRAM controller for parallel processor architecture |
US7974192B2 (en) * | 1999-10-13 | 2011-07-05 | Avaya Inc. | Multicast switching in a distributed communication system |
US7197556B1 (en) * | 1999-10-22 | 2007-03-27 | Nomadix, Inc. | Location-based identification for use in a communications network |
US6934292B1 (en) * | 1999-11-09 | 2005-08-23 | Intel Corporation | Method and system for emulating a single router in a switch stack |
US6618373B1 (en) * | 1999-11-10 | 2003-09-09 | Cisco Technology, Inc. | Method and system for reliable in-order distribution of events |
US6788647B1 (en) | 1999-11-19 | 2004-09-07 | Cisco Technology, Inc. | Automatically applying bi-directional quality of service treatment to network data flows |
US6697380B1 (en) * | 1999-12-07 | 2004-02-24 | Advanced Micro Devices, Inc. | Multiple key lookup arrangement for a shared switching logic address table in a network switch |
US6985964B1 (en) * | 1999-12-22 | 2006-01-10 | Cisco Technology, Inc. | Network processor system including a central processor and at least one peripheral processor |
US6532509B1 (en) | 1999-12-22 | 2003-03-11 | Intel Corporation | Arbitrating command requests in a parallel multi-threaded processing system |
US6694380B1 (en) | 1999-12-27 | 2004-02-17 | Intel Corporation | Mapping requests from a processing unit that uses memory-mapped input-output space |
US6661794B1 (en) | 1999-12-29 | 2003-12-09 | Intel Corporation | Method and apparatus for gigabit packet assignment for multithreaded packet processing |
US7480706B1 (en) * | 1999-12-30 | 2009-01-20 | Intel Corporation | Multi-threaded round-robin receive for fast network port |
US6952824B1 (en) * | 1999-12-30 | 2005-10-04 | Intel Corporation | Multi-threaded sequenced receive for fast network port stream of packets |
US6731644B1 (en) | 2000-02-14 | 2004-05-04 | Cisco Technology, Inc. | Flexible DMA engine for packet header modification |
US6778546B1 (en) | 2000-02-14 | 2004-08-17 | Cisco Technology, Inc. | High-speed hardware implementation of MDRR algorithm over a large number of queues |
US6721316B1 (en) * | 2000-02-14 | 2004-04-13 | Cisco Technology, Inc. | Flexible engine and data structure for packet header processing |
US6813243B1 (en) | 2000-02-14 | 2004-11-02 | Cisco Technology, Inc. | High-speed hardware implementation of red congestion control algorithm |
US6977930B1 (en) | 2000-02-14 | 2005-12-20 | Cisco Technology, Inc. | Pipelined packet switching and queuing architecture |
DE10011667C2 (en) * | 2000-03-10 | 2002-11-21 | Infineon Technologies Ag | High Speed Router |
US7113487B1 (en) * | 2000-04-07 | 2006-09-26 | Intel Corporation | Forwarding control protocol |
US6981027B1 (en) * | 2000-04-10 | 2005-12-27 | International Business Machines Corporation | Method and system for memory management in a network processing system |
US6798777B1 (en) | 2000-04-17 | 2004-09-28 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US7333489B1 (en) * | 2000-05-08 | 2008-02-19 | Crossroads Systems, Inc. | System and method for storing frame header data |
US6807183B1 (en) * | 2000-05-09 | 2004-10-19 | Advanced Micro Devices, Inc. | Arrangement for reading a prescribed location of a FIFO buffer in a network switch port |
DE10022764A1 (en) * | 2000-05-10 | 2001-11-15 | Siemens Ag | Switching system, especially digital switching system, has improved message distributor with internal bus for direct connection of signaling control unit to coupling network |
US7111163B1 (en) | 2000-07-10 | 2006-09-19 | Alterwan, Inc. | Wide area network using internet with quality of service |
US6959332B1 (en) | 2000-07-12 | 2005-10-25 | Cisco Technology, Inc. | Basic command representation of quality of service policies |
US6954462B1 (en) * | 2000-07-31 | 2005-10-11 | Cisco Technology, Inc. | Method and apparatus for determining a multilayer switching path |
US7099932B1 (en) | 2000-08-16 | 2006-08-29 | Cisco Technology, Inc. | Method and apparatus for retrieving network quality of service policy information from a directory in a quality of service policy management system |
US6822940B1 (en) | 2000-09-29 | 2004-11-23 | Cisco Technology, Inc. | Method and apparatus for adapting enforcement of network quality of service policies based on feedback about network conditions |
US7096260B1 (en) | 2000-09-29 | 2006-08-22 | Cisco Technology, Inc. | Marking network data packets with differentiated services codepoints based on network load |
US6988133B1 (en) | 2000-10-31 | 2006-01-17 | Cisco Technology, Inc. | Method and apparatus for communicating network quality of service policy information to a plurality of policy enforcement points |
US7356030B2 (en) | 2000-11-17 | 2008-04-08 | Foundry Networks, Inc. | Network switch cross point |
US7596139B2 (en) | 2000-11-17 | 2009-09-29 | Foundry Networks, Inc. | Backplane interface adapter with error control and redundant fabric |
US6735218B2 (en) * | 2000-11-17 | 2004-05-11 | Foundry Networks, Inc. | Method and system for encoding wide striped cells |
US7236490B2 (en) | 2000-11-17 | 2007-06-26 | Foundry Networks, Inc. | Backplane interface adapter |
US8185615B1 (en) | 2000-11-28 | 2012-05-22 | Verizon Business Global Llc | Message, control and reporting interface for a distributed network access system |
US7046680B1 (en) * | 2000-11-28 | 2006-05-16 | Mci, Inc. | Network access system including a programmable access device having distributed service control |
US7657628B1 (en) | 2000-11-28 | 2010-02-02 | Verizon Business Global Llc | External processor for a distributed network access system |
US8180870B1 (en) * | 2000-11-28 | 2012-05-15 | Verizon Business Global Llc | Programmable access device for a distributed network access system |
US7050396B1 (en) | 2000-11-30 | 2006-05-23 | Cisco Technology, Inc. | Method and apparatus for automatically establishing bi-directional differentiated services treatment of flows in a network |
US7002980B1 (en) | 2000-12-19 | 2006-02-21 | Chiaro Networks, Ltd. | System and method for router queue and congestion management |
KR20020055287A (en) * | 2000-12-28 | 2002-07-08 | 구자홍 | Method for routing a packet of a router device |
US6963535B2 (en) * | 2000-12-28 | 2005-11-08 | Intel Corporation | MAC bus interface |
US6847645B1 (en) * | 2001-02-22 | 2005-01-25 | Cisco Technology, Inc. | Method and apparatus for controlling packet header buffer wrap around in a forwarding engine of an intermediate network node |
US7286532B1 (en) * | 2001-02-22 | 2007-10-23 | Cisco Technology, Inc. | High performance interface logic architecture of an intermediate network node |
US6987760B2 (en) * | 2001-03-05 | 2006-01-17 | International Business Machines Corporation | High speed network processor |
US6909716B1 (en) * | 2001-03-30 | 2005-06-21 | Intel Corporation | System and method for switching and routing data associated with a subscriber community |
US7627870B1 (en) * | 2001-04-28 | 2009-12-01 | Cisco Technology, Inc. | Method and apparatus for a data structure comprising a hierarchy of queues or linked list data structures |
US7206283B2 (en) * | 2001-05-15 | 2007-04-17 | Foundry Networks, Inc. | High-performance network switch |
US7197042B2 (en) * | 2001-06-01 | 2007-03-27 | 4198638 Canada Inc. | Cell-based switch fabric with cell-to-line-card control for regulating injection of packets |
US6744652B2 (en) * | 2001-08-22 | 2004-06-01 | Netlogic Microsystems, Inc. | Concurrent searching of different tables within a content addressable memory |
US20030046429A1 (en) * | 2001-08-30 | 2003-03-06 | Sonksen Bradley Stephen | Static data item processing |
US7139818B1 (en) | 2001-10-04 | 2006-11-21 | Cisco Technology, Inc. | Techniques for dynamic host configuration without direct communications between client and server |
US8543681B2 (en) * | 2001-10-15 | 2013-09-24 | Volli Polymer Gmbh Llc | Network topology discovery systems and methods |
US8868715B2 (en) * | 2001-10-15 | 2014-10-21 | Volli Polymer Gmbh Llc | Report generation and visualization systems and methods and their use in testing frameworks for determining suitability of a network for target applications |
US7447197B2 (en) * | 2001-10-18 | 2008-11-04 | Qlogic, Corporation | System and method of providing network node services |
US7200144B2 (en) * | 2001-10-18 | 2007-04-03 | Qlogic, Corp. | Router and methods using network addresses for virtualization |
US7210003B2 (en) * | 2001-10-31 | 2007-04-24 | Netlogic Microsystems, Inc. | Comparand generation in a content addressable memory |
US6993622B2 (en) * | 2001-10-31 | 2006-01-31 | Netlogic Microsystems, Inc. | Bit level programming interface in a content addressable memory |
GB2382960B (en) * | 2001-12-05 | 2005-03-16 | Ipwireless Inc | Method and arrangement for data processing in a communication system |
US7237058B2 (en) | 2002-01-14 | 2007-06-26 | Netlogic Microsystems, Inc. | Input data selection for content addressable memory |
US7333432B1 (en) | 2002-02-12 | 2008-02-19 | Cisco Technology, Inc. | Method and apparatus for configuring network elements to support real time applications |
US7477600B1 (en) | 2002-02-12 | 2009-01-13 | Cisco Technology, Inc. | Method and apparatus for configuring network elements to support real time applications based on meta-templates |
US7187687B1 (en) | 2002-05-06 | 2007-03-06 | Foundry Networks, Inc. | Pipeline method and system for switching packets |
US20120155466A1 (en) | 2002-05-06 | 2012-06-21 | Ian Edward Davis | Method and apparatus for efficiently processing data packets in a computer network |
US7649885B1 (en) | 2002-05-06 | 2010-01-19 | Foundry Networks, Inc. | Network routing system for enhanced efficiency and monitoring capability |
US7468975B1 (en) | 2002-05-06 | 2008-12-23 | Foundry Networks, Inc. | Flexible method for processing data packets in a network routing system for enhanced efficiency and monitoring capability |
US7266117B1 (en) | 2002-05-06 | 2007-09-04 | Foundry Networks, Inc. | System architecture for very fast ethernet blade |
TW561740B (en) * | 2002-06-06 | 2003-11-11 | Via Tech Inc | Network connecting device and data packet transferring method |
JP4023281B2 (en) * | 2002-10-11 | 2007-12-19 | 株式会社日立製作所 | Packet communication apparatus and packet switch |
US20080008202A1 (en) * | 2002-10-31 | 2008-01-10 | Terrell William C | Router with routing processors and methods for virtualization |
US7107391B2 (en) * | 2002-12-30 | 2006-09-12 | Micron Technology, Inc. | Automatic learning in a CAM |
US7062582B1 (en) * | 2003-03-14 | 2006-06-13 | Marvell International Ltd. | Method and apparatus for bus arbitration dynamic priority based on waiting period |
JP4157403B2 (en) * | 2003-03-19 | 2008-10-01 | 株式会社日立製作所 | Packet communication device |
US6901072B1 (en) | 2003-05-15 | 2005-05-31 | Foundry Networks, Inc. | System and method for high speed packet transmission implementing dual transmit and receive pipelines |
US20060018142A1 (en) * | 2003-08-11 | 2006-01-26 | Varadarajan Srinivasan | Concurrent searching of different tables within a content addressable memory |
US8345701B1 (en) * | 2003-08-26 | 2013-01-01 | F5 Networks, Inc. | Memory system for controlling distribution of packet data across a switch |
US7082493B1 (en) * | 2003-10-31 | 2006-07-25 | Integrated Device Technology, Inc. | CAM-based search engines and packet coprocessors having results status signaling for completed contexts |
US7450592B2 (en) | 2003-11-12 | 2008-11-11 | At&T Intellectual Property I, L.P. | Layer 2/layer 3 interworking via internal virtual UNI |
JP4231773B2 (en) * | 2003-12-01 | 2009-03-04 | 株式会社日立コミュニケーションテクノロジー | VRRP technology that maintains the confidentiality of VR |
US7974275B2 (en) * | 2004-01-09 | 2011-07-05 | Broadcom Corporation | Saturated datagram aging mechanism |
US7817659B2 (en) | 2004-03-26 | 2010-10-19 | Foundry Networks, Llc | Method and apparatus for aggregating input data streams |
US8730961B1 (en) | 2004-04-26 | 2014-05-20 | Foundry Networks, Llc | System and method for optimizing router lookup |
US7853676B1 (en) | 2004-06-10 | 2010-12-14 | Cisco Technology, Inc. | Protocol for efficient exchange of XML documents with a network device |
US8090806B1 (en) | 2004-06-10 | 2012-01-03 | Cisco Technology, Inc. | Two-stage network device configuration process |
US7660882B2 (en) * | 2004-06-10 | 2010-02-09 | Cisco Technology, Inc. | Deploying network element management system provisioning services |
US7657703B1 (en) | 2004-10-29 | 2010-02-02 | Foundry Networks, Inc. | Double density content addressable memory (CAM) lookup scheme |
US20060140191A1 (en) * | 2004-12-29 | 2006-06-29 | Naik Uday R | Multi-level scheduling using single bit vector |
US7751421B2 (en) * | 2004-12-29 | 2010-07-06 | Alcatel Lucent | Traffic generator and monitor |
US7768932B2 (en) * | 2005-04-13 | 2010-08-03 | Hewlett-Packard Development Company, L.P. | Method for analyzing a system in a network |
KR100814904B1 (en) * | 2005-12-06 | 2008-03-19 | 한국전자통신연구원 | On-Chip Communication architecture |
US8054830B2 (en) * | 2005-12-07 | 2011-11-08 | Alcatel Lucent | Managing the distribution of control protocol information in a network node |
US8448162B2 (en) | 2005-12-28 | 2013-05-21 | Foundry Networks, Llc | Hitless software upgrades |
US7903654B2 (en) | 2006-08-22 | 2011-03-08 | Foundry Networks, Llc | System and method for ECMP load sharing |
US8238255B2 (en) | 2006-11-22 | 2012-08-07 | Foundry Networks, Llc | Recovering from failures without impact on data traffic in a shared bus architecture |
US7978614B2 (en) | 2007-01-11 | 2011-07-12 | Foundry Network, LLC | Techniques for detecting non-receipt of fault detection protocol packets |
US8271859B2 (en) | 2007-07-18 | 2012-09-18 | Foundry Networks Llc | Segmented CRC design in high speed networks |
US8037399B2 (en) | 2007-07-18 | 2011-10-11 | Foundry Networks, Llc | Techniques for segmented CRC design in high speed networks |
US8149839B1 (en) | 2007-09-26 | 2012-04-03 | Foundry Networks, Llc | Selection of trunk ports and paths using rotation |
US8190881B2 (en) * | 2007-10-15 | 2012-05-29 | Foundry Networks Llc | Scalable distributed web-based authentication |
US8606940B2 (en) * | 2008-02-06 | 2013-12-10 | Alcatel Lucent | DHCP address conflict detection/enforcement |
US8335238B2 (en) * | 2008-12-23 | 2012-12-18 | International Business Machines Corporation | Reassembling streaming data across multiple packetized communication channels |
US8176026B2 (en) * | 2009-04-14 | 2012-05-08 | International Business Machines Corporation | Consolidating file system backend operations with access of data |
US8266504B2 (en) | 2009-04-14 | 2012-09-11 | International Business Machines Corporation | Dynamic monitoring of ability to reassemble streaming data across multiple channels based on history |
US8271634B2 (en) * | 2009-04-30 | 2012-09-18 | Alcatel Lucent | Buffer system for managing service measurement requests |
US8090901B2 (en) | 2009-05-14 | 2012-01-03 | Brocade Communications Systems, Inc. | TCAM management approach that minimize movements |
US8971335B2 (en) * | 2009-07-02 | 2015-03-03 | Exafer Ltd | System and method for creating a transitive optimized flow path |
US8325733B2 (en) * | 2009-09-09 | 2012-12-04 | Exafer Ltd | Method and system for layer 2 manipulator and forwarder |
US8599850B2 (en) | 2009-09-21 | 2013-12-03 | Brocade Communications Systems, Inc. | Provisioning single or multistage networks using ethernet service instances (ESIs) |
US8274987B2 (en) * | 2010-03-22 | 2012-09-25 | International Business Machines Corporation | Contention free pipelined broadcasting within a constant bisection bandwidth network topology |
US10250528B2 (en) * | 2012-11-13 | 2019-04-02 | Netronome Systems, Inc. | Packet prediction in a multi-protocol label switching network using operation, administration, and maintenance (OAM) messaging |
US10834065B1 (en) | 2015-03-31 | 2020-11-10 | F5 Networks, Inc. | Methods for SSL protected NTLM re-authentication and devices thereof |
US10404698B1 (en) | 2016-01-15 | 2019-09-03 | F5 Networks, Inc. | Methods for adaptive organization of web application access points in webtops and devices thereof |
Family Cites Families (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US4821265A (en) | 1987-04-06 | 1989-04-11 | Racal Data Communications Inc. | Node architecture for communication networks |
US4949338A (en) | 1987-04-06 | 1990-08-14 | Racal Data Communications Inc. | Arbitration in multiprocessor communication node |
DE4106183A1 (en) | 1991-02-27 | 1992-09-03 | Standard Elektrik Lorenz Ag | DEVICE AND METHOD FOR D-CHANNEL PACKAGING |
US5295133A (en) * | 1992-02-12 | 1994-03-15 | Sprint International Communications Corp. | System administration in a flat distributed packet switch architecture |
US5926482A (en) * | 1994-05-05 | 1999-07-20 | Sprint Communications Co. L.P. | Telecommunications apparatus, system, and method with an enhanced signal transfer point |
US5805072A (en) * | 1994-12-12 | 1998-09-08 | Ultra-High Speed Network | VC connection method |
US5600644A (en) * | 1995-03-10 | 1997-02-04 | At&T | Method and apparatus for interconnecting LANs |
US5581552A (en) * | 1995-05-23 | 1996-12-03 | At&T | Multimedia server |
CA2181293C (en) * | 1995-07-17 | 2000-06-06 | Charles Kevin Huscroft | Atm layer device |
US6067608A (en) * | 1997-04-15 | 2000-05-23 | Bull Hn Information Systems Inc. | High performance mechanism for managing allocation of virtual memory buffers to virtual processes on a least recently used basis |
US5909686A (en) * | 1997-06-30 | 1999-06-01 | Sun Microsystems, Inc. | Hardware-assisted central processing unit access to a forwarding database |
US5938736A (en) * | 1997-06-30 | 1999-08-17 | Sun Microsystems, Inc. | Search engine architecture for a high performance multi-layer switch element |
US6021132A (en) * | 1997-06-30 | 2000-02-01 | Sun Microsystems, Inc. | Shared memory management in a switched network element |
US6016310A (en) * | 1997-06-30 | 2000-01-18 | Sun Microsystems, Inc. | Trunking support in a high performance network device |
-
1998
- 1998-07-17 US US09/118,458 patent/US6424659B2/en not_active Expired - Lifetime
Cited By (364)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US7688324B1 (en) * | 1999-03-05 | 2010-03-30 | Zoran Corporation | Interactive set-top box having a unified memory architecture |
US20050141501A1 (en) * | 1999-03-17 | 2005-06-30 | Broadcom Corporation | Network switch having a programmable counter |
US7720055B2 (en) * | 1999-03-17 | 2010-05-18 | Broadcom Corporation | Method for handling IP multicast packets in network switch |
US7782891B2 (en) | 1999-03-17 | 2010-08-24 | Broadcom Corporation | Network switch memory interface configuration |
US6859459B1 (en) * | 1999-11-16 | 2005-02-22 | Nec Corporation | High-speed/high-reliability ether transmission system and I/F apparatus |
US6934260B1 (en) * | 2000-02-01 | 2005-08-23 | Advanced Micro Devices, Inc. | Arrangement for controlling learning of layer 3 network addresses in a network switch |
US6925641B1 (en) * | 2000-02-04 | 2005-08-02 | Xronix Communications, Inc. | Real time DSP load management system |
US6674769B1 (en) * | 2000-03-07 | 2004-01-06 | Advanced Micro Devices, Inc. | Simultaneous searching of layer 3 policy filter and policy cache in a network switch port |
US8503304B2 (en) | 2000-04-17 | 2013-08-06 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US8879395B2 (en) | 2000-04-17 | 2014-11-04 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US8238246B2 (en) | 2000-04-17 | 2012-08-07 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US9258228B2 (en) | 2000-04-17 | 2016-02-09 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US7986629B1 (en) | 2000-04-17 | 2011-07-26 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US7215637B1 (en) * | 2000-04-17 | 2007-05-08 | Juniper Networks, Inc. | Systems and methods for processing packets |
US9813339B2 (en) | 2000-04-17 | 2017-11-07 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US7688727B1 (en) | 2000-04-17 | 2010-03-30 | Juniper Networks, Inc. | Filtering and route lookup in a switching device |
US6757742B1 (en) * | 2000-05-25 | 2004-06-29 | Advanced Micro Devices, Inc. | Computer-based system for validating hash-based table lookup schemes in a network switch |
US6798778B1 (en) * | 2000-06-14 | 2004-09-28 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for updating context information for a core processor |
US6826180B1 (en) * | 2000-06-14 | 2004-11-30 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for storing summation blocks of context information for a core processor |
US6845099B2 (en) * | 2000-06-14 | 2005-01-18 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for modifying selectors to retrieve context information for a core processor |
US6791983B1 (en) * | 2000-06-14 | 2004-09-14 | Mindspeed Technologies, Inc. | Content-addressable memory for use with a communication packet processor to retrieve context information |
US6891829B1 (en) * | 2000-06-14 | 2005-05-10 | Mindspeed Technologies, Inc. | Communication packet processor with a look-up engine and content-addressable memory for retrieving context information for a core processor |
US8027341B2 (en) * | 2000-06-23 | 2011-09-27 | Broadcom Corporation | Switch having external address resolution interface |
US20060256787A1 (en) * | 2000-06-23 | 2006-11-16 | Broadcom Corporation | Switch having external address resolution interface |
US6885666B1 (en) * | 2000-08-14 | 2005-04-26 | Advanced Micro Devices, Inc. | Apparatus and method in a network switch for synchronizing transfer of a control tag to a switch fabric with transfer of frame data to a buffer memory |
US7894427B2 (en) * | 2000-09-12 | 2011-02-22 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US20110103387A1 (en) * | 2000-09-12 | 2011-05-05 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US20060120366A1 (en) * | 2000-09-12 | 2006-06-08 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US9042381B2 (en) | 2000-09-12 | 2015-05-26 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US8675650B2 (en) | 2000-09-12 | 2014-03-18 | Cisco Technology, Inc. | Stateful network address translation protocol implemented over a data network |
US7680137B2 (en) * | 2000-12-12 | 2010-03-16 | Airbus France S.A.S. | Process for transmitting asynchronous data packets |
US20040114600A1 (en) * | 2000-12-12 | 2004-06-17 | Jean-Pierre Mao | Process for transmitting asynchronous data packets |
US7590134B2 (en) | 2000-12-12 | 2009-09-15 | Airbus France S.A.S. | System for transmitting asynchronous data packets |
US7586928B2 (en) * | 2000-12-12 | 2009-09-08 | Airbus France | Process and device for deterministic transmission of asynchronous data in packets |
US20020105958A1 (en) * | 2000-12-12 | 2002-08-08 | Jean-Pierre Mao | Process and device for deterministic transmission of asynchronous data in packets |
US6778526B1 (en) * | 2000-12-22 | 2004-08-17 | Nortel Networks Limited | High speed access bus interface and protocol |
US20020120888A1 (en) * | 2001-02-14 | 2002-08-29 | Jorg Franke | Network co-processor for vehicles |
US7260668B2 (en) * | 2001-02-14 | 2007-08-21 | Micronas Gmbh | Network co-processor for vehicles |
US6940854B1 (en) * | 2001-03-23 | 2005-09-06 | Advanced Micro Devices, Inc. | Systems and methods for determining priority based on previous priority determinations |
US20020143910A1 (en) * | 2001-03-29 | 2002-10-03 | Shih-Wei Chou | Network hub |
US7136935B2 (en) * | 2001-06-22 | 2006-11-14 | Inktomi Corporation | Efficient data transmissions based on a policy |
US20030007488A1 (en) * | 2001-06-22 | 2003-01-09 | Sriram Rao | Efficient data transmission based on a policy |
US7216204B2 (en) * | 2001-08-27 | 2007-05-08 | Intel Corporation | Mechanism for providing early coherency detection to enable high performance memory updates in a latency sensitive multithreaded environment |
US20030041216A1 (en) * | 2001-08-27 | 2003-02-27 | Rosenbluth Mark B. | Mechanism for providing early coherency detection to enable high performance memory updates in a latency sensitive multithreaded environment |
US20030095567A1 (en) * | 2001-11-20 | 2003-05-22 | Lo Man Kuk | Real time protocol packet handler |
US7280527B2 (en) | 2002-05-13 | 2007-10-09 | International Business Machines Corporation | Logically grouping physical ports into logical interfaces to expand bandwidth |
US20030210688A1 (en) * | 2002-05-13 | 2003-11-13 | International Business Machines Corporation | Logically grouping physical ports into logical interfaces to expand bandwidth |
EP1363429A2 (en) * | 2002-05-13 | 2003-11-19 | Alcatel | Logically grouping physical ports into logical interfaces to expand bandwidth |
EP1363429A3 (en) * | 2002-05-13 | 2004-06-23 | Alcatel | Logically grouping physical ports into logical interfaces to expand bandwidth |
US7181567B2 (en) * | 2002-06-04 | 2007-02-20 | Lucent Technologies Inc. | Hitless restart of access control module |
US20030225965A1 (en) * | 2002-06-04 | 2003-12-04 | Ram Krishnan | Hitless restart of access control module |
US7689485B2 (en) * | 2002-08-10 | 2010-03-30 | Cisco Technology, Inc. | Generating accounting data based on access control list entries |
US20040172346A1 (en) * | 2002-08-10 | 2004-09-02 | Cisco Technology, Inc., A California Corporation | Generating accounting data based on access control list entries |
US20040139290A1 (en) * | 2003-01-10 | 2004-07-15 | Gilbert Wolrich | Memory interleaving |
US20050185437A1 (en) * | 2003-01-10 | 2005-08-25 | Gilbert Wolrich | Memory interleaving |
US20110110372A1 (en) * | 2003-06-05 | 2011-05-12 | Juniper Networks, Inc. | Systems and methods to perform hybrid switching and routing functions |
US20100153961A1 (en) * | 2004-02-10 | 2010-06-17 | Hitachi, Ltd. | Storage system having processor and interface adapters that can be increased or decreased based on required performance |
US20050198429A1 (en) * | 2004-03-02 | 2005-09-08 | Nec Electronics Corporation | Multilayer system and clock control method |
US20060114895A1 (en) * | 2004-11-30 | 2006-06-01 | Broadcom Corporation | CPU transmission of unmodified packets |
US8170019B2 (en) * | 2004-11-30 | 2012-05-01 | Broadcom Corporation | CPU transmission of unmodified packets |
US20070118677A1 (en) * | 2005-05-13 | 2007-05-24 | Freescale Semiconductor Incorporated | Packet switch having a crossbar switch that connects multiport receiving and transmitting elements |
US20070136331A1 (en) * | 2005-11-28 | 2007-06-14 | Nec Laboratories America | Storage-efficient and collision-free hash-based packet processing architecture and method |
US7653670B2 (en) * | 2005-11-28 | 2010-01-26 | Nec Laboratories America, Inc. | Storage-efficient and collision-free hash-based packet processing architecture and method |
US9900410B2 (en) | 2006-05-01 | 2018-02-20 | Nicira, Inc. | Private ethernet overlay networks over a shared ethernet in a virtual environment |
US8611350B2 (en) * | 2006-07-21 | 2013-12-17 | Cortina Systems, Inc. | Apparatus and method for layer-2 to 7 search engine for high speed network application |
US20110080913A1 (en) * | 2006-07-21 | 2011-04-07 | Cortina Systems, Inc. | Apparatus and method for layer-2 to 7 search engine for high speed network application |
US20080126572A1 (en) * | 2006-10-05 | 2008-05-29 | Holt John M | Multi-path switching networks |
US20080144785A1 (en) * | 2006-12-19 | 2008-06-19 | Dae-Hyun Lee | Call setup method and terminal in a IP network |
US8600017B2 (en) * | 2006-12-19 | 2013-12-03 | Samsung Electronics Co., Ltd. | Call setup method and terminal in an IP network |
US20080225837A1 (en) * | 2007-03-16 | 2008-09-18 | Novell, Inc. | System and Method for Multi-Layer Distributed Switching |
US20090037629A1 (en) * | 2007-08-01 | 2009-02-05 | Broadcom Corporation | Master slave core architecture with direct buses |
US9876672B2 (en) | 2007-09-26 | 2018-01-23 | Nicira, Inc. | Network operating system for managing and securing networks |
US11683214B2 (en) | 2007-09-26 | 2023-06-20 | Nicira, Inc. | Network operating system for managing and securing networks |
US10749736B2 (en) | 2007-09-26 | 2020-08-18 | Nicira, Inc. | Network operating system for managing and securing networks |
US9083609B2 (en) | 2007-09-26 | 2015-07-14 | Nicira, Inc. | Network operating system for managing and securing networks |
US11757797B2 (en) | 2008-05-23 | 2023-09-12 | Vmware, Inc. | Distributed virtual switch for virtualized computer systems |
US11190463B2 (en) | 2008-05-23 | 2021-11-30 | Vmware, Inc. | Distributed virtual switch for virtualized computer systems |
US9559987B1 (en) * | 2008-09-26 | 2017-01-31 | Tellabs Operations, Inc | Method and apparatus for improving CAM learn throughput using a cache |
US10681136B2 (en) | 2009-02-19 | 2020-06-09 | Micron Technology, Inc. | Memory network methods, apparatus, and systems |
US20140032701A1 (en) * | 2009-02-19 | 2014-01-30 | Micron Technology, Inc. | Memory network methods, apparatus, and systems |
US9590919B2 (en) | 2009-04-01 | 2017-03-07 | Nicira, Inc. | Method and apparatus for implementing and managing virtual switches |
US11425055B2 (en) | 2009-04-01 | 2022-08-23 | Nicira, Inc. | Method and apparatus for implementing and managing virtual switches |
US8966035B2 (en) | 2009-04-01 | 2015-02-24 | Nicira, Inc. | Method and apparatus for implementing and managing distributed virtual switches in several hosts and physical forwarding elements |
US10931600B2 (en) | 2009-04-01 | 2021-02-23 | Nicira, Inc. | Method and apparatus for implementing and managing virtual switches |
US9306910B2 (en) | 2009-07-27 | 2016-04-05 | Vmware, Inc. | Private allocated networks over shared communications infrastructure |
US9697032B2 (en) | 2009-07-27 | 2017-07-04 | Vmware, Inc. | Automated network configuration of virtual machines in a virtual lab environment |
US10949246B2 (en) | 2009-07-27 | 2021-03-16 | Vmware, Inc. | Automated network configuration of virtual machines in a virtual lab environment |
US9952892B2 (en) | 2009-07-27 | 2018-04-24 | Nicira, Inc. | Automated network configuration of virtual machines in a virtual lab environment |
US10757234B2 (en) | 2009-09-30 | 2020-08-25 | Nicira, Inc. | Private allocated networks over shared communications infrastructure |
US9888097B2 (en) | 2009-09-30 | 2018-02-06 | Nicira, Inc. | Private allocated networks over shared communications infrastructure |
US11533389B2 (en) | 2009-09-30 | 2022-12-20 | Nicira, Inc. | Private allocated networks over shared communications infrastructure |
US10291753B2 (en) | 2009-09-30 | 2019-05-14 | Nicira, Inc. | Private allocated networks over shared communications infrastructure |
US11917044B2 (en) | 2009-09-30 | 2024-02-27 | Nicira, Inc. | Private allocated networks over shared communications infrastructure |
US11838395B2 (en) | 2010-06-21 | 2023-12-05 | Nicira, Inc. | Private ethernet overlay networks over a shared ethernet in a virtual environment |
US10951744B2 (en) | 2010-06-21 | 2021-03-16 | Nicira, Inc. | Private ethernet overlay networks over a shared ethernet in a virtual environment |
US8880468B2 (en) | 2010-07-06 | 2014-11-04 | Nicira, Inc. | Secondary storage architecture for a network control system that utilizes a primary network information base |
US9680750B2 (en) | 2010-07-06 | 2017-06-13 | Nicira, Inc. | Use of tunnels to hide network addresses |
US9007903B2 (en) | 2010-07-06 | 2015-04-14 | Nicira, Inc. | Managing a network by controlling edge and non-edge switching elements |
US9008087B2 (en) | 2010-07-06 | 2015-04-14 | Nicira, Inc. | Processing requests in a network control system with multiple controller instances |
US11743123B2 (en) | 2010-07-06 | 2023-08-29 | Nicira, Inc. | Managed switch architectures: software managed switches, hardware managed switches, and heterogeneous managed switches |
US8964528B2 (en) | 2010-07-06 | 2015-02-24 | Nicira, Inc. | Method and apparatus for robust packet distribution among hierarchical managed switching elements |
US10038597B2 (en) | 2010-07-06 | 2018-07-31 | Nicira, Inc. | Mesh architectures for managed switching elements |
US9049153B2 (en) | 2010-07-06 | 2015-06-02 | Nicira, Inc. | Logical packet processing pipeline that retains state information to effectuate efficient processing of packets |
US10103939B2 (en) | 2010-07-06 | 2018-10-16 | Nicira, Inc. | Network control apparatus and method for populating logical datapath sets |
US9077664B2 (en) | 2010-07-06 | 2015-07-07 | Nicira, Inc. | One-hop packet processing in a network with managed switching elements |
US8966040B2 (en) | 2010-07-06 | 2015-02-24 | Nicira, Inc. | Use of network information base structure to establish communication between applications |
US9106587B2 (en) | 2010-07-06 | 2015-08-11 | Nicira, Inc. | Distributed network control system with one master controller per managed switching element |
US9112811B2 (en) | 2010-07-06 | 2015-08-18 | Nicira, Inc. | Managed switching elements used as extenders |
US9172663B2 (en) | 2010-07-06 | 2015-10-27 | Nicira, Inc. | Method and apparatus for replicating network information base in a distributed network control system with multiple controller instances |
US8717895B2 (en) | 2010-07-06 | 2014-05-06 | Nicira, Inc. | Network virtualization apparatus and method with a table mapping engine |
US8718070B2 (en) | 2010-07-06 | 2014-05-06 | Nicira, Inc. | Distributed network virtualization apparatus and method |
US9231891B2 (en) | 2010-07-06 | 2016-01-05 | Nicira, Inc. | Deployment of hierarchical managed switching elements |
US10320585B2 (en) | 2010-07-06 | 2019-06-11 | Nicira, Inc. | Network control apparatus and method for creating and modifying logical switching elements |
US10326660B2 (en) | 2010-07-06 | 2019-06-18 | Nicira, Inc. | Network virtualization apparatus and method |
US9300603B2 (en) | 2010-07-06 | 2016-03-29 | Nicira, Inc. | Use of rich context tags in logical data processing |
US11876679B2 (en) | 2010-07-06 | 2024-01-16 | Nicira, Inc. | Method and apparatus for interacting with a network information base in a distributed network control system with multiple controller instances |
US8958292B2 (en) | 2010-07-06 | 2015-02-17 | Nicira, Inc. | Network control apparatus and method with port security controls |
US8959215B2 (en) | 2010-07-06 | 2015-02-17 | Nicira, Inc. | Network virtualization |
US9306875B2 (en) | 2010-07-06 | 2016-04-05 | Nicira, Inc. | Managed switch architectures for implementing logical datapath sets |
US11677588B2 (en) | 2010-07-06 | 2023-06-13 | Nicira, Inc. | Network control apparatus and method for creating and modifying logical switching elements |
US8743888B2 (en) | 2010-07-06 | 2014-06-03 | Nicira, Inc. | Network control apparatus and method |
US8743889B2 (en) | 2010-07-06 | 2014-06-03 | Nicira, Inc. | Method and apparatus for using a network information base to control a plurality of shared network infrastructure switching elements |
US8750119B2 (en) | 2010-07-06 | 2014-06-10 | Nicira, Inc. | Network control apparatus and method with table mapping engine |
US10686663B2 (en) | 2010-07-06 | 2020-06-16 | Nicira, Inc. | Managed switch architectures: software managed switches, hardware managed switches, and heterogeneous managed switches |
US8750164B2 (en) | 2010-07-06 | 2014-06-10 | Nicira, Inc. | Hierarchical managed switch architecture |
US8761036B2 (en) | 2010-07-06 | 2014-06-24 | Nicira, Inc. | Network control apparatus and method with quality of service controls |
US9363210B2 (en) | 2010-07-06 | 2016-06-07 | Nicira, Inc. | Distributed network control system with one master controller per logical datapath set |
US9692655B2 (en) | 2010-07-06 | 2017-06-27 | Nicira, Inc. | Packet processing in a network with hierarchical managed switching elements |
US8964598B2 (en) | 2010-07-06 | 2015-02-24 | Nicira, Inc. | Mesh architectures for managed switching elements |
US9391928B2 (en) | 2010-07-06 | 2016-07-12 | Nicira, Inc. | Method and apparatus for interacting with a network information base in a distributed network control system with multiple controller instances |
US8775594B2 (en) | 2010-07-06 | 2014-07-08 | Nicira, Inc. | Distributed network control system with a distributed hash table |
US11641321B2 (en) | 2010-07-06 | 2023-05-02 | Nicira, Inc. | Packet processing for logical datapath sets |
US11539591B2 (en) | 2010-07-06 | 2022-12-27 | Nicira, Inc. | Distributed network control system with one master controller per logical datapath set |
US8913483B2 (en) | 2010-07-06 | 2014-12-16 | Nicira, Inc. | Fault tolerant managed switching element architecture |
US8817620B2 (en) | 2010-07-06 | 2014-08-26 | Nicira, Inc. | Network virtualization apparatus and method |
US8817621B2 (en) | 2010-07-06 | 2014-08-26 | Nicira, Inc. | Network virtualization apparatus |
US11509564B2 (en) | 2010-07-06 | 2022-11-22 | Nicira, Inc. | Method and apparatus for replicating network information base in a distributed network control system with multiple controller instances |
US8830823B2 (en) | 2010-07-06 | 2014-09-09 | Nicira, Inc. | Distributed control platform for large-scale production networks |
US9525647B2 (en) | 2010-07-06 | 2016-12-20 | Nicira, Inc. | Network control apparatus and method for creating and modifying logical switching elements |
US11223531B2 (en) | 2010-07-06 | 2022-01-11 | Nicira, Inc. | Method and apparatus for interacting with a network information base in a distributed network control system with multiple controller instances |
US10021019B2 (en) | 2010-07-06 | 2018-07-10 | Nicira, Inc. | Packet processing for logical datapath sets |
US8837493B2 (en) | 2010-07-06 | 2014-09-16 | Nicira, Inc. | Distributed network control apparatus and method |
US8842679B2 (en) | 2010-07-06 | 2014-09-23 | Nicira, Inc. | Control system that elects a master controller instance for switching elements |
US8856419B2 (en) | 2010-07-19 | 2014-10-07 | International Business Machines Corporation | Register access in distributed virtual bridge environment |
US9043452B2 (en) | 2011-05-04 | 2015-05-26 | Nicira, Inc. | Network control apparatus and method for port isolation |
US9185069B2 (en) | 2011-08-17 | 2015-11-10 | Nicira, Inc. | Handling reverse NAT in logical L3 routing |
US9444651B2 (en) | 2011-08-17 | 2016-09-13 | Nicira, Inc. | Flow generation from second level controller to first level controller to managed switching element |
US11804987B2 (en) | 2011-08-17 | 2023-10-31 | Nicira, Inc. | Flow generation from second level controller to first level controller to managed switching element |
WO2013026049A1 (en) * | 2011-08-17 | 2013-02-21 | Nicira, Inc. | Distributed logical l3 routing |
US10868761B2 (en) | 2011-08-17 | 2020-12-15 | Nicira, Inc. | Logical L3 daemon |
US9407599B2 (en) | 2011-08-17 | 2016-08-02 | Nicira, Inc. | Handling NAT migration in logical L3 routing |
US9059999B2 (en) | 2011-08-17 | 2015-06-16 | Nicira, Inc. | Load balancing in a logical pipeline |
US9369426B2 (en) | 2011-08-17 | 2016-06-14 | Nicira, Inc. | Distributed logical L3 routing |
US9356906B2 (en) | 2011-08-17 | 2016-05-31 | Nicira, Inc. | Logical L3 routing with DHCP |
US9350696B2 (en) | 2011-08-17 | 2016-05-24 | Nicira, Inc. | Handling NAT in logical L3 routing |
US10193708B2 (en) | 2011-08-17 | 2019-01-29 | Nicira, Inc. | Multi-domain interconnect |
US9319375B2 (en) | 2011-08-17 | 2016-04-19 | Nicira, Inc. | Flow templating in logical L3 routing |
US10027584B2 (en) | 2011-08-17 | 2018-07-17 | Nicira, Inc. | Distributed logical L3 routing |
US10931481B2 (en) | 2011-08-17 | 2021-02-23 | Nicira, Inc. | Multi-domain interconnect |
US9461960B2 (en) | 2011-08-17 | 2016-10-04 | Nicira, Inc. | Logical L3 daemon |
US11695695B2 (en) | 2011-08-17 | 2023-07-04 | Nicira, Inc. | Logical L3 daemon |
US9276897B2 (en) | 2011-08-17 | 2016-03-01 | Nicira, Inc. | Distributed logical L3 routing |
US8958298B2 (en) | 2011-08-17 | 2015-02-17 | Nicira, Inc. | Centralized logical L3 routing |
US9306864B2 (en) | 2011-10-25 | 2016-04-05 | Nicira, Inc. | Scheduling distribution of physical control plane data |
US9602421B2 (en) | 2011-10-25 | 2017-03-21 | Nicira, Inc. | Nesting transaction updates to minimize communication |
US9319337B2 (en) | 2011-10-25 | 2016-04-19 | Nicira, Inc. | Universal physical control plane |
US11669488B2 (en) | 2011-10-25 | 2023-06-06 | Nicira, Inc. | Chassis controller |
US9319338B2 (en) | 2011-10-25 | 2016-04-19 | Nicira, Inc. | Tunnel creation |
US9954793B2 (en) | 2011-10-25 | 2018-04-24 | Nicira, Inc. | Chassis controller |
US10505856B2 (en) | 2011-10-25 | 2019-12-10 | Nicira, Inc. | Chassis controller |
US9319336B2 (en) | 2011-10-25 | 2016-04-19 | Nicira, Inc. | Scheduling distribution of logical control plane data |
US10884780B2 (en) | 2011-11-15 | 2021-01-05 | Nicira, Inc. | Architecture of networks with middleboxes |
US10191763B2 (en) | 2011-11-15 | 2019-01-29 | Nicira, Inc. | Architecture of networks with middleboxes |
US10922124B2 (en) | 2011-11-15 | 2021-02-16 | Nicira, Inc. | Network control system for configuring middleboxes |
US10310886B2 (en) | 2011-11-15 | 2019-06-04 | Nicira, Inc. | Network control system for configuring middleboxes |
US11593148B2 (en) | 2011-11-15 | 2023-02-28 | Nicira, Inc. | Network control system for configuring middleboxes |
US10977067B2 (en) | 2011-11-15 | 2021-04-13 | Nicira, Inc. | Control plane interface for logical middlebox services |
US10089127B2 (en) | 2011-11-15 | 2018-10-02 | Nicira, Inc. | Control plane interface for logical middlebox services |
US10235199B2 (en) | 2011-11-15 | 2019-03-19 | Nicira, Inc. | Migrating middlebox state for distributed middleboxes |
US11372671B2 (en) | 2011-11-15 | 2022-06-28 | Nicira, Inc. | Architecture of networks with middleboxes |
US10514941B2 (en) | 2011-11-15 | 2019-12-24 | Nicira, Inc. | Load balancing and destination network address translation middleboxes |
US11740923B2 (en) | 2011-11-15 | 2023-08-29 | Nicira, Inc. | Architecture of networks with middleboxes |
US10949248B2 (en) | 2011-11-15 | 2021-03-16 | Nicira, Inc. | Load balancing and destination network address translation middleboxes |
US20130182707A1 (en) * | 2012-01-18 | 2013-07-18 | International Business Machines Corporation | Managing a global forwarding table in a distributed switch |
US8861400B2 (en) | 2012-01-18 | 2014-10-14 | International Business Machines Corporation | Requesting multicast membership information in a distributed switch in response to a miss event |
US8891535B2 (en) * | 2012-01-18 | 2014-11-18 | International Business Machines Corporation | Managing a global forwarding table in a distributed switch |
US10135676B2 (en) | 2012-04-18 | 2018-11-20 | Nicira, Inc. | Using transactions to minimize churn in a distributed network control system |
US10033579B2 (en) | 2012-04-18 | 2018-07-24 | Nicira, Inc. | Using transactions to compute and propagate network forwarding state |
US9306845B2 (en) * | 2012-11-22 | 2016-04-05 | Hitachi Metals, Ltd. | Communication system and network relay device |
US20140140346A1 (en) * | 2012-11-22 | 2014-05-22 | Hitachi Metals, Ltd. | Communication System and Network Relay Device |
US20140351885A1 (en) * | 2013-05-22 | 2014-11-27 | Unisys Corporation | Control of simple network management protocol activity |
US9038136B2 (en) * | 2013-05-22 | 2015-05-19 | Unisys Corporation | Control of simple network management protocol activity |
US10680948B2 (en) | 2013-07-08 | 2020-06-09 | Nicira, Inc. | Hybrid packet processing |
US10033640B2 (en) | 2013-07-08 | 2018-07-24 | Nicira, Inc. | Hybrid packet processing |
US9571386B2 (en) | 2013-07-08 | 2017-02-14 | Nicira, Inc. | Hybrid packet processing |
US10181993B2 (en) | 2013-07-12 | 2019-01-15 | Nicira, Inc. | Tracing network packets through logical and physical networks |
US10778557B2 (en) | 2013-07-12 | 2020-09-15 | Nicira, Inc. | Tracing network packets through logical and physical networks |
US9407580B2 (en) | 2013-07-12 | 2016-08-02 | Nicira, Inc. | Maintaining data stored with a packet |
US11201808B2 (en) | 2013-07-12 | 2021-12-14 | Nicira, Inc. | Tracing logical network packets through physical network |
US10764238B2 (en) | 2013-08-14 | 2020-09-01 | Nicira, Inc. | Providing services for logical networks |
US11695730B2 (en) | 2013-08-14 | 2023-07-04 | Nicira, Inc. | Providing services for logical networks |
US9887960B2 (en) | 2013-08-14 | 2018-02-06 | Nicira, Inc. | Providing services for logical networks |
US9952885B2 (en) | 2013-08-14 | 2018-04-24 | Nicira, Inc. | Generation of configuration files for a DHCP module executing within a virtualized container |
US9577845B2 (en) | 2013-09-04 | 2017-02-21 | Nicira, Inc. | Multiple active L3 gateways for logical networks |
US10003534B2 (en) | 2013-09-04 | 2018-06-19 | Nicira, Inc. | Multiple active L3 gateways for logical networks |
US10389634B2 (en) | 2013-09-04 | 2019-08-20 | Nicira, Inc. | Multiple active L3 gateways for logical networks |
US9503371B2 (en) | 2013-09-04 | 2016-11-22 | Nicira, Inc. | High availability L3 gateways for logical networks |
US10382324B2 (en) | 2013-09-15 | 2019-08-13 | Nicira, Inc. | Dynamically generating flows with wildcard fields |
US9602398B2 (en) | 2013-09-15 | 2017-03-21 | Nicira, Inc. | Dynamically generating flows with wildcard fields |
US10498638B2 (en) | 2013-09-15 | 2019-12-03 | Nicira, Inc. | Performing a multi-stage lookup to classify packets |
US20170070474A1 (en) * | 2013-10-06 | 2017-03-09 | Mellanox Technologies Ltd. | Simplified packet routing |
US10708219B2 (en) * | 2013-10-06 | 2020-07-07 | Mellanox Technologies, Ltd. | Simplified packet routing |
US10063458B2 (en) | 2013-10-13 | 2018-08-28 | Nicira, Inc. | Asymmetric connection with external networks |
US10528373B2 (en) | 2013-10-13 | 2020-01-07 | Nicira, Inc. | Configuration of logical router |
US9575782B2 (en) | 2013-10-13 | 2017-02-21 | Nicira, Inc. | ARP for logical router |
US9910686B2 (en) | 2013-10-13 | 2018-03-06 | Nicira, Inc. | Bridging between network segments with a logical router |
US9785455B2 (en) | 2013-10-13 | 2017-10-10 | Nicira, Inc. | Logical router |
US10693763B2 (en) | 2013-10-13 | 2020-06-23 | Nicira, Inc. | Asymmetric connection with external networks |
US11029982B2 (en) | 2013-10-13 | 2021-06-08 | Nicira, Inc. | Configuration of logical router |
US9977685B2 (en) | 2013-10-13 | 2018-05-22 | Nicira, Inc. | Configuration of logical router |
US10193771B2 (en) | 2013-12-09 | 2019-01-29 | Nicira, Inc. | Detecting and handling elephant flows |
US11811669B2 (en) | 2013-12-09 | 2023-11-07 | Nicira, Inc. | Inspecting operations of a machine to detect elephant flows |
US9967199B2 (en) | 2013-12-09 | 2018-05-08 | Nicira, Inc. | Inspecting operations of a machine to detect elephant flows |
US9548924B2 (en) | 2013-12-09 | 2017-01-17 | Nicira, Inc. | Detecting an elephant flow based on the size of a packet |
US10158538B2 (en) | 2013-12-09 | 2018-12-18 | Nicira, Inc. | Reporting elephant flows to a network controller |
US10666530B2 (en) | 2013-12-09 | 2020-05-26 | Nicira, Inc | Detecting and handling large flows |
US11539630B2 (en) | 2013-12-09 | 2022-12-27 | Nicira, Inc. | Inspecting operations of a machine to detect elephant flows |
US11095536B2 (en) | 2013-12-09 | 2021-08-17 | Nicira, Inc. | Detecting and handling large flows |
US9838276B2 (en) | 2013-12-09 | 2017-12-05 | Nicira, Inc. | Detecting an elephant flow based on the size of a packet |
US9569368B2 (en) | 2013-12-13 | 2017-02-14 | Nicira, Inc. | Installing and managing flows in a flow table cache |
US10380019B2 (en) | 2013-12-13 | 2019-08-13 | Nicira, Inc. | Dynamically adjusting the number of flows allowed in a flow table cache |
US9996467B2 (en) | 2013-12-13 | 2018-06-12 | Nicira, Inc. | Dynamically adjusting the number of flows allowed in a flow table cache |
US10567283B2 (en) | 2014-03-14 | 2020-02-18 | Nicira, Inc. | Route advertisement by managed gateways |
US9590901B2 (en) | 2014-03-14 | 2017-03-07 | Nicira, Inc. | Route advertisement by managed gateways |
US9419855B2 (en) | 2014-03-14 | 2016-08-16 | Nicira, Inc. | Static routes for logical routers |
US10164881B2 (en) | 2014-03-14 | 2018-12-25 | Nicira, Inc. | Route advertisement by managed gateways |
US10110431B2 (en) | 2014-03-14 | 2018-10-23 | Nicira, Inc. | Logical router processing by network controller |
US9313129B2 (en) | 2014-03-14 | 2016-04-12 | Nicira, Inc. | Logical router processing by network controller |
US11025543B2 (en) | 2014-03-14 | 2021-06-01 | Nicira, Inc. | Route advertisement by managed gateways |
US9225597B2 (en) | 2014-03-14 | 2015-12-29 | Nicira, Inc. | Managed gateways peering with external router to attract ingress packets |
US10411955B2 (en) | 2014-03-21 | 2019-09-10 | Nicira, Inc. | Multiple levels of logical routers |
US9647883B2 (en) | 2014-03-21 | 2017-05-09 | Nicria, Inc. | Multiple levels of logical routers |
US11252024B2 (en) | 2014-03-21 | 2022-02-15 | Nicira, Inc. | Multiple levels of logical routers |
US9503321B2 (en) | 2014-03-21 | 2016-11-22 | Nicira, Inc. | Dynamic routing for logical routers |
US11190443B2 (en) | 2014-03-27 | 2021-11-30 | Nicira, Inc. | Address resolution using multiple designated instances of a logical router |
US11736394B2 (en) | 2014-03-27 | 2023-08-22 | Nicira, Inc. | Address resolution using multiple designated instances of a logical router |
US9893988B2 (en) | 2014-03-27 | 2018-02-13 | Nicira, Inc. | Address resolution using multiple designated instances of a logical router |
US9413644B2 (en) | 2014-03-27 | 2016-08-09 | Nicira, Inc. | Ingress ECMP in virtual distributed routing environment |
US10659373B2 (en) | 2014-03-31 | 2020-05-19 | Nicira, Inc | Processing packets according to hierarchy of flow entry storages |
US10193806B2 (en) | 2014-03-31 | 2019-01-29 | Nicira, Inc. | Performing a finishing operation to improve the quality of a resulting hash |
US11431639B2 (en) | 2014-03-31 | 2022-08-30 | Nicira, Inc. | Caching of service decisions |
US9385954B2 (en) | 2014-03-31 | 2016-07-05 | Nicira, Inc. | Hashing techniques for use in a network environment |
AU2014398480B2 (en) * | 2014-06-24 | 2018-04-05 | Hitachi, Ltd. | Financial products trading system and financial products trading control method |
US10540717B2 (en) | 2014-06-24 | 2020-01-21 | Hitachi, Ltd. | Financial products trading system and financial products trading control method |
US9742881B2 (en) | 2014-06-30 | 2017-08-22 | Nicira, Inc. | Network virtualization using just-in-time distributed capability for classification encoding |
US11252037B2 (en) | 2014-09-30 | 2022-02-15 | Nicira, Inc. | Using physical location to modify behavior of a distributed virtual network element |
US9768980B2 (en) | 2014-09-30 | 2017-09-19 | Nicira, Inc. | Virtual distributed bridging |
US10250443B2 (en) | 2014-09-30 | 2019-04-02 | Nicira, Inc. | Using physical location to modify behavior of a distributed virtual network element |
US10511458B2 (en) | 2014-09-30 | 2019-12-17 | Nicira, Inc. | Virtual distributed bridging |
US11178051B2 (en) | 2014-09-30 | 2021-11-16 | Vmware, Inc. | Packet key parser for flow-based forwarding elements |
US10020960B2 (en) | 2014-09-30 | 2018-07-10 | Nicira, Inc. | Virtual distributed bridging |
US11483175B2 (en) | 2014-09-30 | 2022-10-25 | Nicira, Inc. | Virtual distributed bridging |
US10469342B2 (en) | 2014-10-10 | 2019-11-05 | Nicira, Inc. | Logical network traffic analysis |
US11128550B2 (en) | 2014-10-10 | 2021-09-21 | Nicira, Inc. | Logical network traffic analysis |
US11283731B2 (en) | 2015-01-30 | 2022-03-22 | Nicira, Inc. | Logical router with multiple routing components |
US10129180B2 (en) | 2015-01-30 | 2018-11-13 | Nicira, Inc. | Transit logical switch within logical router |
US11799800B2 (en) | 2015-01-30 | 2023-10-24 | Nicira, Inc. | Logical router with multiple routing components |
US10079779B2 (en) | 2015-01-30 | 2018-09-18 | Nicira, Inc. | Implementing logical router uplinks |
US10700996B2 (en) | 2015-01-30 | 2020-06-30 | Nicira, Inc | Logical router with multiple routing components |
US10652143B2 (en) | 2015-04-04 | 2020-05-12 | Nicira, Inc | Route server mode for dynamic routing between logical and physical networks |
US10038628B2 (en) | 2015-04-04 | 2018-07-31 | Nicira, Inc. | Route server mode for dynamic routing between logical and physical networks |
US11601362B2 (en) | 2015-04-04 | 2023-03-07 | Nicira, Inc. | Route server mode for dynamic routing between logical and physical networks |
US9923760B2 (en) | 2015-04-06 | 2018-03-20 | Nicira, Inc. | Reduction of churn in a network control system |
US9967134B2 (en) | 2015-04-06 | 2018-05-08 | Nicira, Inc. | Reduction of network churn based on differences in input state |
US11799775B2 (en) | 2015-06-30 | 2023-10-24 | Nicira, Inc. | Intermediate logical interfaces in a virtual distributed router environment |
US10348625B2 (en) | 2015-06-30 | 2019-07-09 | Nicira, Inc. | Sharing common L2 segment in a virtual distributed router environment |
US10225184B2 (en) | 2015-06-30 | 2019-03-05 | Nicira, Inc. | Redirecting traffic in a virtual distributed router environment |
US10693783B2 (en) | 2015-06-30 | 2020-06-23 | Nicira, Inc. | Intermediate logical interfaces in a virtual distributed router environment |
US11050666B2 (en) | 2015-06-30 | 2021-06-29 | Nicira, Inc. | Intermediate logical interfaces in a virtual distributed router environment |
US10361952B2 (en) | 2015-06-30 | 2019-07-23 | Nicira, Inc. | Intermediate logical interfaces in a virtual distributed router environment |
US11533256B2 (en) | 2015-08-11 | 2022-12-20 | Nicira, Inc. | Static route configuration for logical router |
US10230629B2 (en) | 2015-08-11 | 2019-03-12 | Nicira, Inc. | Static route configuration for logical router |
US10805212B2 (en) | 2015-08-11 | 2020-10-13 | Nicira, Inc. | Static route configuration for logical router |
US10129142B2 (en) | 2015-08-11 | 2018-11-13 | Nicira, Inc. | Route configuration for logical router |
US11425021B2 (en) | 2015-08-31 | 2022-08-23 | Nicira, Inc. | Authorization for advertised routes among logical routers |
US10601700B2 (en) | 2015-08-31 | 2020-03-24 | Nicira, Inc. | Authorization for advertised routes among logical routers |
US10057157B2 (en) | 2015-08-31 | 2018-08-21 | Nicira, Inc. | Automatically advertising NAT routes between logical routers |
US10075363B2 (en) | 2015-08-31 | 2018-09-11 | Nicira, Inc. | Authorization for advertised routes among logical routers |
US10204122B2 (en) | 2015-09-30 | 2019-02-12 | Nicira, Inc. | Implementing an interface between tuple and message-driven control entities |
US11288249B2 (en) | 2015-09-30 | 2022-03-29 | Nicira, Inc. | Implementing an interface between tuple and message-driven control entities |
US11593145B2 (en) | 2015-10-31 | 2023-02-28 | Nicira, Inc. | Static route types for logical routers |
US10795716B2 (en) | 2015-10-31 | 2020-10-06 | Nicira, Inc. | Static route types for logical routers |
US10095535B2 (en) | 2015-10-31 | 2018-10-09 | Nicira, Inc. | Static route types for logical routers |
US10819621B2 (en) | 2016-02-23 | 2020-10-27 | Mellanox Technologies Tlv Ltd. | Unicast forwarding of adaptive-routing notifications |
US11502958B2 (en) | 2016-04-28 | 2022-11-15 | Nicira, Inc. | Automatic configuration of logical routers on edge nodes |
US10805220B2 (en) | 2016-04-28 | 2020-10-13 | Nicira, Inc. | Automatic configuration of logical routers on edge nodes |
US10333849B2 (en) | 2016-04-28 | 2019-06-25 | Nicira, Inc. | Automatic configuration of logical routers on edge nodes |
US10841273B2 (en) | 2016-04-29 | 2020-11-17 | Nicira, Inc. | Implementing logical DHCP servers in logical networks |
US11855959B2 (en) | 2016-04-29 | 2023-12-26 | Nicira, Inc. | Implementing logical DHCP servers in logical networks |
US11019167B2 (en) | 2016-04-29 | 2021-05-25 | Nicira, Inc. | Management of update queues for network controller |
US11601521B2 (en) | 2016-04-29 | 2023-03-07 | Nicira, Inc. | Management of update queues for network controller |
US10484515B2 (en) | 2016-04-29 | 2019-11-19 | Nicira, Inc. | Implementing logical metadata proxy servers in logical networks |
US10091161B2 (en) | 2016-04-30 | 2018-10-02 | Nicira, Inc. | Assignment of router ID for logical routers |
US10178029B2 (en) | 2016-05-11 | 2019-01-08 | Mellanox Technologies Tlv Ltd. | Forwarding of adaptive routing notifications |
US10749801B2 (en) | 2016-06-29 | 2020-08-18 | Nicira, Inc. | Installation of routing tables for logical router in route server mode |
US11418445B2 (en) | 2016-06-29 | 2022-08-16 | Nicira, Inc. | Installation of routing tables for logical router in route server mode |
US10560320B2 (en) | 2016-06-29 | 2020-02-11 | Nicira, Inc. | Ranking of gateways in cluster |
US10153973B2 (en) | 2016-06-29 | 2018-12-11 | Nicira, Inc. | Installation of routing tables for logical router in route server mode |
US11539574B2 (en) | 2016-08-31 | 2022-12-27 | Nicira, Inc. | Edge node cluster network redundancy and fast convergence using an underlay anycast VTEP IP |
US10454758B2 (en) | 2016-08-31 | 2019-10-22 | Nicira, Inc. | Edge node cluster network redundancy and fast convergence using an underlay anycast VTEP IP |
US10341236B2 (en) | 2016-09-30 | 2019-07-02 | Nicira, Inc. | Anycast edge service gateways |
US10911360B2 (en) | 2016-09-30 | 2021-02-02 | Nicira, Inc. | Anycast edge service gateways |
US10237123B2 (en) | 2016-12-21 | 2019-03-19 | Nicira, Inc. | Dynamic recovery from a split-brain failure in edge nodes |
US10645204B2 (en) | 2016-12-21 | 2020-05-05 | Nicira, Inc | Dynamic recovery from a split-brain failure in edge nodes |
US10742746B2 (en) | 2016-12-21 | 2020-08-11 | Nicira, Inc. | Bypassing a load balancer in a return path of network traffic |
US10212071B2 (en) | 2016-12-21 | 2019-02-19 | Nicira, Inc. | Bypassing a load balancer in a return path of network traffic |
US11665242B2 (en) | 2016-12-21 | 2023-05-30 | Nicira, Inc. | Bypassing a load balancer in a return path of network traffic |
US10616045B2 (en) | 2016-12-22 | 2020-04-07 | Nicira, Inc. | Migration of centralized routing components of logical router |
US10200294B2 (en) | 2016-12-22 | 2019-02-05 | Mellanox Technologies Tlv Ltd. | Adaptive routing based on flow-control credits |
US11115262B2 (en) | 2016-12-22 | 2021-09-07 | Nicira, Inc. | Migration of centralized routing components of logical router |
US10805239B2 (en) | 2017-03-07 | 2020-10-13 | Nicira, Inc. | Visualization of path between logical network endpoints |
US11336590B2 (en) | 2017-03-07 | 2022-05-17 | Nicira, Inc. | Visualization of path between logical network endpoints |
US10200306B2 (en) | 2017-03-07 | 2019-02-05 | Nicira, Inc. | Visualization of packet tracing operation results |
US20190007364A1 (en) * | 2017-06-30 | 2019-01-03 | Nicira, Inc. | Assignment of unique physical network addresses for logical network addresses |
US10637800B2 (en) | 2017-06-30 | 2020-04-28 | Nicira, Inc | Replacement of logical network addresses with physical network addresses |
US11595345B2 (en) | 2017-06-30 | 2023-02-28 | Nicira, Inc. | Assignment of unique physical network addresses for logical network addresses |
US10681000B2 (en) * | 2017-06-30 | 2020-06-09 | Nicira, Inc. | Assignment of unique physical network addresses for logical network addresses |
US20190065493A1 (en) * | 2017-08-28 | 2019-02-28 | International Business Machines Corporation | Efficient and accurate lookups of data by a stream processor using a hash table |
US20190065494A1 (en) * | 2017-08-28 | 2019-02-28 | International Business Machines Corporation | Efficient and accurate lookups of data by a stream processor using a hash table |
US10803040B2 (en) * | 2017-08-28 | 2020-10-13 | International Business Machines Corporation | Efficient and accurate lookups of data by a stream processor using a hash table |
US10817491B2 (en) * | 2017-08-28 | 2020-10-27 | International Business Machines Corporation | Efficient and accurate lookups of data by a stream processor using a hash table |
US10608887B2 (en) | 2017-10-06 | 2020-03-31 | Nicira, Inc. | Using packet tracing tool to automatically execute packet capture operations |
US11336486B2 (en) | 2017-11-14 | 2022-05-17 | Nicira, Inc. | Selection of managed forwarding element for bridge spanning multiple datacenters |
US10374827B2 (en) | 2017-11-14 | 2019-08-06 | Nicira, Inc. | Identifier that maps to different networks at different datacenters |
US10511459B2 (en) | 2017-11-14 | 2019-12-17 | Nicira, Inc. | Selection of managed forwarding element for bridge spanning multiple datacenters |
US10644995B2 (en) | 2018-02-14 | 2020-05-05 | Mellanox Technologies Tlv Ltd. | Adaptive routing in a box |
US11855898B1 (en) * | 2018-03-14 | 2023-12-26 | F5, Inc. | Methods for traffic dependent direct memory access optimization and devices thereof |
US10931560B2 (en) | 2018-11-23 | 2021-02-23 | Vmware, Inc. | Using route type to determine routing protocol behavior |
US10797998B2 (en) | 2018-12-05 | 2020-10-06 | Vmware, Inc. | Route server for distributed routers using hierarchical routing protocol |
US10938788B2 (en) | 2018-12-12 | 2021-03-02 | Vmware, Inc. | Static routes for policy-based VPN |
US11005724B1 (en) | 2019-01-06 | 2021-05-11 | Mellanox Technologies, Ltd. | Network topology having minimal number of long connections among groups of network elements |
US11159343B2 (en) | 2019-08-30 | 2021-10-26 | Vmware, Inc. | Configuring traffic optimization using distributed edge services |
US11095480B2 (en) | 2019-08-30 | 2021-08-17 | Vmware, Inc. | Traffic optimization using distributed edge services |
US11924080B2 (en) | 2020-01-17 | 2024-03-05 | VMware LLC | Practical overlay network latency measurement in datacenter |
US11606294B2 (en) | 2020-07-16 | 2023-03-14 | Vmware, Inc. | Host computer configured to facilitate distributed SNAT service |
US11616755B2 (en) | 2020-07-16 | 2023-03-28 | Vmware, Inc. | Facilitating distributed SNAT service |
US11611613B2 (en) | 2020-07-24 | 2023-03-21 | Vmware, Inc. | Policy-based forwarding to a load balancer of a load balancing cluster |
US11902050B2 (en) | 2020-07-28 | 2024-02-13 | VMware LLC | Method for providing distributed gateway service at host computer |
US11451413B2 (en) | 2020-07-28 | 2022-09-20 | Vmware, Inc. | Method for advertising availability of distributed gateway service and machines at host computer |
US11196628B1 (en) | 2020-07-29 | 2021-12-07 | Vmware, Inc. | Monitoring container clusters |
US11558426B2 (en) | 2020-07-29 | 2023-01-17 | Vmware, Inc. | Connection tracking for container cluster |
US11570090B2 (en) | 2020-07-29 | 2023-01-31 | Vmware, Inc. | Flow tracing operation in container cluster |
US11575594B2 (en) | 2020-09-10 | 2023-02-07 | Mellanox Technologies, Ltd. | Deadlock-free rerouting for resolving local link failures using detour paths |
US11411911B2 (en) | 2020-10-26 | 2022-08-09 | Mellanox Technologies, Ltd. | Routing across multiple subnetworks using address mapping |
US11736436B2 (en) | 2020-12-31 | 2023-08-22 | Vmware, Inc. | Identifying routes with indirect addressing in a datacenter |
US11336533B1 (en) | 2021-01-08 | 2022-05-17 | Vmware, Inc. | Network visualization of correlations between logical elements and associated physical elements |
US11848825B2 (en) | 2021-01-08 | 2023-12-19 | Vmware, Inc. | Network visualization of correlations between logical elements and associated physical elements |
US11870682B2 (en) | 2021-06-22 | 2024-01-09 | Mellanox Technologies, Ltd. | Deadlock-free local rerouting for handling multiple local link failures in hierarchical network topologies |
US11687210B2 (en) | 2021-07-05 | 2023-06-27 | Vmware, Inc. | Criteria-based expansion of group nodes in a network topology visualization |
US11711278B2 (en) | 2021-07-24 | 2023-07-25 | Vmware, Inc. | Visualization of flow trace operation across multiple sites |
US11855862B2 (en) | 2021-09-17 | 2023-12-26 | Vmware, Inc. | Tagging packets for monitoring and analysis |
US11706109B2 (en) | 2021-09-17 | 2023-07-18 | Vmware, Inc. | Performance of traffic monitoring actions |
US11677645B2 (en) | 2021-09-17 | 2023-06-13 | Vmware, Inc. | Traffic monitoring |
US11765103B2 (en) | 2021-12-01 | 2023-09-19 | Mellanox Technologies, Ltd. | Large-scale network with high port utilization |
Also Published As
Publication number | Publication date |
---|---|
US6424659B2 (en) | 2002-07-23 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
US6424659B2 (en) | Multi-layer switching apparatus and method | |
US7065050B1 (en) | Apparatus and method for controlling data flow in a network switch | |
US7114024B2 (en) | Apparatus and method for managing memory defects | |
JP3806239B2 (en) | Network switch with shared memory system | |
US6201789B1 (en) | Network switch with dynamic backpressure per port | |
US6430626B1 (en) | Network switch with a multiple bus structure and a bridge interface for transferring network data between different buses | |
US6460088B1 (en) | Method and apparatus for port vector determination at egress | |
US6504846B1 (en) | Method and apparatus for reclaiming buffers using a single buffer bit | |
JP3789395B2 (en) | Packet processing device | |
US6389480B1 (en) | Programmable arbitration system for determining priority of the ports of a network switch | |
US6260073B1 (en) | Network switch including a switch manager for periodically polling the network ports to determine their status and controlling the flow of data between ports | |
US6052751A (en) | Method and apparatus for changing the number of access slots into a memory | |
JP4068166B2 (en) | Search engine architecture for high performance multilayer switch elements | |
EP0854615A2 (en) | Multiport polling system for a network switch | |
JPH10233797A (en) | Network switch provided with statistics read access | |
JPH10215266A (en) | Network switch with another cut-through buffer | |
JPH10243013A (en) | Method and device executing simultaneous read write cycle by network switch | |
WO2000072524A1 (en) | Apparatus and method for programmable memory access slot assignment | |
US6907036B1 (en) | Network switch enhancements directed to processing of internal operations in the network switch | |
US6778547B1 (en) | Method and apparatus for improving throughput of a rules checker logic | |
US6335938B1 (en) | Multiport communication switch having gigaport and expansion ports sharing the same time slot in internal rules checker | |
US6895015B1 (en) | Dynamic time slot allocation in internal rules checker scheduler | |
US6816488B1 (en) | Apparatus and method for processing data frames in a network switch | |
US6480490B1 (en) | Interleaved access to address table in network switching system | |
US7031302B1 (en) | High-speed stats gathering in a network switch |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
AS | Assignment |
Owner name: FLOWWISE NETWORKS INC., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:VEERINA, MAHESH;REEL/FRAME:009850/0588 Effective date: 19980922 |
|
AS | Assignment |
Owner name: FLOWWISE NETWORKS INC., CALIFORNIA Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:VISWANADHAM, KRISHNA;REEL/FRAME:009851/0253 Effective date: 19980717 |
|
AS | Assignment |
Owner name: NETWORK EQUIPMENT TECHNOLOGIES, INC., CALIFORNIA Free format text: MERGER;ASSIGNOR:FLOWWISE NETWORKS, INC.;REEL/FRAME:011186/0804 Effective date: 19991214 |
|
STCF | Information on status: patent grant |
Free format text: PATENTED CASE |
|
FEPP | Fee payment procedure |
Free format text: PAT HOLDER NO LONGER CLAIMS SMALL ENTITY STATUS, ENTITY STATUS SET TO UNDISCOUNTED (ORIGINAL EVENT CODE: STOL); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
REFU | Refund |
Free format text: REFUND - SURCHARGE, PETITION TO ACCEPT PYMT AFTER EXP, UNINTENTIONAL (ORIGINAL EVENT CODE: R2551); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 4 |
|
FPAY | Fee payment |
Year of fee payment: 8 |
|
FEPP | Fee payment procedure |
Free format text: PAYOR NUMBER ASSIGNED (ORIGINAL EVENT CODE: ASPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY Free format text: PAYER NUMBER DE-ASSIGNED (ORIGINAL EVENT CODE: RMPN); ENTITY STATUS OF PATENT OWNER: LARGE ENTITY |
|
FPAY | Fee payment |
Year of fee payment: 12 |
|
AS | Assignment |
Owner name: BANK OF AMERICA, N.A., AS ADMINISTRATIVE AGENT, MASSACHUSETTS Free format text: SECURITY INTEREST;ASSIGNORS:SONUS NETWORKS, INC.;SONUS FEDERAL, INC.;NETWORK EQUIPMENT TECHNOLOGIES, INC.;AND OTHERS;REEL/FRAME:033728/0409 Effective date: 20140627 Owner name: BANK OF AMERICA, N.A., AS ADMINISTRATIVE AGENT, MA Free format text: SECURITY INTEREST;ASSIGNORS:SONUS NETWORKS, INC.;SONUS FEDERAL, INC.;NETWORK EQUIPMENT TECHNOLOGIES, INC.;AND OTHERS;REEL/FRAME:033728/0409 Effective date: 20140627 |
|
AS | Assignment |
Owner name: TAQUA, INC., TEXAS Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 Owner name: SONUS NETWORKS, INC., MASSACHUSETTS Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 Owner name: SONUS INTERNATIONAL, INC., MASSACHUSETTS Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 Owner name: SONUS FEDERAL, INC., VIRGINIA Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 Owner name: PERFORMANCE TECHNOLOGIES, INCORPORATED, NEW YORK Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 Owner name: NETWORK EQUIPMENT TECHNOLOGIES, INC., CALIFORNIA Free format text: RELEASE BY SECURED PARTY;ASSIGNOR:BANK OF AMERICA, N.A.;REEL/FRAME:044283/0361 Effective date: 20170701 |
|
AS | Assignment |
Owner name: SONUS NETWORKS, INC., MASSACHUSETTS Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:NETWORK EQUIPMENT TECHNOLOGIES, INC.;REEL/FRAME:044904/0829 Effective date: 20171218 |