WO2008074191A1

WO2008074191A1 - A packet switching system and a method thereof

Info

Publication number: WO2008074191A1
Application number: PCT/CN2006/003526
Authority: WO
Inventors: Zufa Yang
Original assignee: Zte Corporation
Priority date: 2006-12-21
Filing date: 2006-12-21
Publication date: 2008-06-26
Also published as: CN101542985B; CN101542985A

Abstract

An equipment for packet switching and a method thereof, the packet switching equipment includes: a packet unit, for dividing or filling the input data into the fixed length packet; a switching unit, for dividing at least one priority for the unicast packet and the multicast packet respectively which are from the packet unit, and respectively assigning a queue to the unicast packet and the multicast packet of each priority, then scheduling it according to the priority of the queue. The unicast packet queue which is assigned to the highest priority and the multicast packet queue which is assigned to the highest priority are used to carrying the TDM service or the packet service, and the unicast packet queue which is assigned to the other priority and the multicast packet queue which is assigned to the other priority are used to carrying the packet service. The invention is also applied to the TDM service and the packet service switching.

Description

TECHNICAL FIELD The present invention relates to the field of communications, and in particular, to a packet switching apparatus and method. BACKGROUND OF THE INVENTION Time Division Multiplexing (TDM) networks are based on fixed time slot cross-connection services, and data of the same service occupies fixed-length time slot forwarding at equal intervals, and is adapted to transmit delay-sensitive services, such as real-time voice and high definition. Video information, etc. The packet switched network is based on the storage forwarding and statistical multiplexing exchange services. The interval between adjacent packets of the same service is uncertain, and it is suitable for transmitting services that are not sensitive to delay, such as emails, files, forbidden images, and the like. The two network characteristics are different, and the bearer services are highly differentiated. At the same time, they are widely used in the telecommunications field. In recent years, people have been exploring ways to simplify the structure of telecommunication networks, reduce the cost of network construction, improve the utilization of existing network resources, and have successively launched support for multi-service provisioning platforms or multi-service provisioning platforms or Multi. -Service Switching Platform (MSPP for short) and Pseudo Wire Emulation Edge-to-Edge (PWE3). The former uses a unified data encapsulation format to simultaneously pass packet services and TDM services through Synchronous Digital Hierarchy (SDH) or Synchronous Optical Network (SONET) or Plesiochronous Digital Hierarchy (PDH). Or Optical Transport Hierarchy (abbreviated as ΟΤΗ) network transmission, the latter uses edge node emulation technology to simultaneously transmit packet services and TDM services over a packet switched network. With the rapid development of the Internet, the share of packet services in telecommunications networks far exceeds that of TDM services, and the gap is still increasing. In line with this, most of the service switching equipment of the telecommunication network adopts packet switching.

TDM services often require switching in byte-interleaved manner, such as TDM services such as El and T1 transmitted over SDH or SONET networks. Different service data is transmitted by byte interleaving and multiplexing. If the time slot of the input direction is extended from single byte to multibyte on the line card of the TDM switching device, The TDM synchronization time slot crossover unit crosses the multi-byte time slot, and the output direction of the online card is restored to the byte interleaving mode, which can realize the equivalent function of the byte interleaving TDM switching device. The usual packet switching components are implemented in a shared store-and-forward and output-queue-based manner. All ports sharing a unified memory will cause the service to compete for the internal buffer resources of the switching device. When multiple service statistics are multiplexed to the same port, the service will compete for the port bandwidth resources. An effective way to solve the competition is to use the queue management buffer to limit the occupation of the shared buffer by each service, and according to a certain strategy, such as scheduling the port < ^ team based on priority. The variable length characteristics of the packet necessarily introduce an indeterminate buffer delay. In the process of packet exchange, the emergence of resource competition will also introduce uncertain forwarding delay. Therefore, the traditional packet switching method is not suitable for transmitting TDM services. The packet switching network and TDM switching network will also be used for a long period of time. coexist. It can be seen from the above that a new packet switching method suitable for transmitting TDM services is needed, so that both TDM services and packet services can be transmitted through the new packet switching manner. SUMMARY OF THE INVENTION In view of the above problems, the present invention proposes a packet switching apparatus and method to simultaneously accommodate the switching requirements of TDM services and packet services. According to an aspect of the present invention, a packet switching apparatus is provided, comprising: a grouping unit 102 for dividing or padding input data into fixed length packets; and an exchange unit 104 for unicasting packets from the grouping unit And the multicast packet are respectively divided into at least one priority and each queue is assigned a unicast packet and a multicast packet for each priority, and is scheduled according to the priority of the queue, wherein the unicast packet is assigned to the highest priority The queue and the queue assigned to the highest priority multicast packet are used to carry TDM traffic or packet traffic, and the queues assigned to other priority unicast packets and the queues assigned to other prior multicast packets are used for bearer. Grouping business. In the above packet switching apparatus, the queue is used to buffer the address of the packet, and operates in a first-in first-out manner. In the above packet switching apparatus, the switching unit includes: a unicast shared buffer for storing unicast packets from the packet unit; a multicast shared buffer for storing multicast packets from the packet unit; a packet scheduling unit, Used to dispatch packets from the queue according to the priority of the queue; queue management unit, used Allocate shared buffer space for each port's unicast packet queue and multicast packet queue, and allocate unicast shared buffer space for the highest priority unicast packet queue for TDM unicast service or unicast Packet service, the multicast shared buffer space allocated for the highest priority multicast packet queue is used for TDM multicast service or multicast packet service, and is released after the unicast packet and the multicast packet are forwarded to the designated port. a shared buffer space corresponding to the unicast packet and the multicast packet; and an address management unit for managing the unicast shared buffer and the idle storage address of the multicast shared buffer period. In the above packet switching apparatus, the write speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet input rates of all the input ports of the switching unit, the unicast shared buffer and the multicast sharing. The read speed of the buffer is not lower than the sum of the packet output speeds of all the output ports of the switching unit. In the above packet switching apparatus, the queue management unit copies the storage address of the multicast packet to the corresponding multicast queue of the plurality of designated ports in the multicast shared buffer, and when the predetermined time is reached, the multicast packet storage address is forwarded. . In the above packet switching apparatus, in the case where the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the occupied multicast is used. In the case where the shared buffer space is lower than the assigned value, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. In the above packet switching apparatus, in the case where the unicast shared buffer has free storage space, it is allowed to buffer the currently input unicast packet, otherwise discard the currently input unicast packet; and there is idle in the multicast shared buffer. In the case of storage space, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. In the above packet switching apparatus, the queue scheduling unit schedules the unicast service and the multicast service of the same priority in the following manner: preferentially scheduling the unicast service, preferentially scheduling the multicast service, and randomly scheduling. In the above packet switching apparatus, the queue scheduling unit preferentially schedules packets in the highest priority queue. According to another aspect of the present invention, a packet switching method is provided, comprising the steps of:

S1202: The grouping unit divides or fills the input data into fixed length packets; and in S1204, the switching unit divides the unicast packet and the multicast packet from the grouping unit into at least one priority and each for each The priority unicast packet and the multicast packet are respectively assigned a queue, and the packet is scheduled according to the priority of the queue, wherein the queue assigned to the highest priority unicast packet and the queue allocated to the highest priority multicast packet are used. For carrying TDM services or packet services, queues assigned to other priority unicast packets and queues allocated to other priority multicast packets are used to carry packet traffic. In the above packet switching method, the queue is used to buffer the address of the packet, and operates in a first-in first-out manner. In the above packet switching method, step S1204 includes: S12042, storing a unicast packet from a packet unit through a unicast shared buffer and using a unicast shared buffer space allocated for the highest priority unicast packet queue for TDM Unicast traffic or unicast packet traffic, and multicast packet buffers from the packet unit through the multicast shared buffer and the multicast shared buffer allocated for the highest priority multicast packet queue for TDM multicast traffic Or multicast packet service; and S 12044, a packet scheduling unit, configured to schedule a packet from the queue according to a priority of the queue; and S12046, after the unicast packet and the multicast packet are forwarded to the designated port, releasing the unicast packet and The shared buffer space corresponding to the multicast packet. In the above packet switching method, the write speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet input rates of all the input ports of the switching unit, the unicast shared buffer and the multicast sharing slow. The readout speed of the punch is not lower than the sum of the packet output speeds of all the output ports of the switching unit. In the above packet switching method, the multicast shared buffer copies the storage address of the multicast packet to the corresponding multicast queue of the plurality of designated ports, and when the predetermined time is reached, releases the multicast packet storage address. In the above packet switching method, in the case that the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the occupied multicast sharing is occupied. In the case where the buffer space is lower than the assigned value, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. In the above packet switching method, in the case where the unicast shared buffer has free storage space, it is allowed to buffer the currently input unicast packet, otherwise discard the currently input unicast packet; and exist in the multicast shared buffer. In the case of free storage space, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. In the foregoing packet switching method, the unicast service and the multicast service of the same priority are scheduled in the following manner: unicast service is preferentially scheduled, multicast service is preferentially scheduled, and random scheduling is performed. The present invention ensures that the highest priority service preferentially occupies the global shared buffer space, and uses the highest priority service packet for forwarding TDM service or packet service, and other priority packets are used only for 4 packets of packet service, thereby making the present invention The packet switching apparatus and method are capable of simultaneously adapting the exchange of TDM services and packet services. BRIEF DESCRIPTION OF THE DRAWINGS The accompanying drawings, which are set to illustrate,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,, In the drawings: FIG. 1 is a functional block diagram of a packet switching apparatus according to an embodiment of the present invention; FIG. 2 is a functional structural diagram of a packet switching apparatus according to an embodiment of the present invention; FIG. 3 is a unicast sharing according to an embodiment of the present invention. FIG. 4 is a diagram of a multicast sharing buffer gate P艮 setting according to an embodiment of the present invention; FIG. 5 is a unicast priority queue structure diagram according to an embodiment of the present invention; A unicast priority queue threshold setting map of an embodiment of the invention; FIG. 7 is a map of a unicast priority queue to a unicast shared buffer according to an embodiment of the present invention; FIG. 8 is a multicast priority according to an embodiment of the present invention. FIG. 9 is a multicast priority queue threshold setting diagram according to an embodiment of the present invention; FIG. 10 is a mapping diagram of a multicast priority queue to a multicast shared buffer according to an embodiment of the present invention; a port queue scheduling time slot distribution map according to an embodiment of the present invention; Figure 12 is a flow chart of a packet switching method in accordance with an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, specific embodiments of the present invention will be described in detail with reference to the accompanying drawings. Referring to Figure 1, a packet switching apparatus in accordance with an embodiment of the present invention is illustrated. As shown in FIG. 1, the packet switching apparatus includes: a grouping unit 102, configured to divide or fill input data into fixed length packets; and an exchanging unit 104, configured to separately separate unicast packets and multicast packets from the packet unit Dividing into at least one priority and assigning a queue to each of the unicast packets and the multicast packets of each priority, and scheduling according to the priority of the queue, wherein the queues assigned to the highest priority unicast packets are The queue assigned to the highest priority multicast packet is used for TDM service or packet service, the queue assigned to other priority unicast packets and the queue allocated to other priority multicast packets are used to carry the packet business. The queue is used to buffer the address of the packet and operates in a first-in, first-out manner. The switching unit 104 includes: a unicast shared buffer 1042 for storing unicast packets from the packet unit; a multicast shared buffer 1044 for storing multicast packets from the packet unit; a packet scheduling unit 1046, A priority is used to schedule a packet from the queue; a queue management unit 1048 is configured to allocate a shared buffer space for each port's unicast packet queue and multicast packet queue, respectively, and will be the highest priority list. The unicast shared buffer space allocated by the broadcast packet queue is used for TDM unicast service or unicast packet service, and the multicast shared buffer space allocated for the highest priority multicast packet queue is used for TDM multicast service or multicast. Packet service, and after the unicast packet and the multicast packet are forwarded to the designated port, releasing the shared buffer space corresponding to the unicast packet and the multicast packet; and an address management unit 1050 for managing the unicast shared buffer and The idle storage address of the multicast shared buffer period. Wherein, the write speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet input rates of all the input ports of the switching unit, and the reading of the unicast shared buffer and the multicast shared buffer The output speed is not lower than the sum of the packet output speeds of all the output ports of the switching unit. The queue management unit copies the storage address of the multicast packet to the corresponding multicast queue of the plurality of designated ports in the multicast shared buffer, and releases the multicast packet storage address when the predetermined time is reached. Wherein, in the case that the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the occupied multicast shared buffer space is lower than In the case of assigning a value, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. Wherein, in the case that the unicast shared buffer has free storage space, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the idle storage space exists in the multicast shared buffer. Next, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. The queue scheduling unit schedules unicast services and multicast services of the same priority in the following manner: preferentially scheduling unicast services, preferentially scheduling multicast services, and scheduling randomly. The queue scheduling unit preferentially schedules packets in the highest priority queue. Specifically, as shown in FIG. 2, in an actual network, a packet switching apparatus according to an embodiment of the present invention includes the following parts: a Unicast Share Buffer (USB), a multicast shared buffer (Multicast). Share Buffer (MSB), Unicast Priority Queue (UPQ), Multicast Priority Queue (MPQ), Packet Scheduler (PSch), Queue Management (Queue) Management, referred to as QM), and Address Management (AM). In the following description, we support the packet switching apparatus of the embodiment of the present invention to divide the unicast service and the multicast service into four priority levels, respectively. The four priorities of the unicast service and the multicast service are named Prio (Priority 0, Priority 0), Pril, Pri2, Pri3, respectively. Among them, PriO has the highest priority, and Pri3 has the highest priority. TDM attributes can be selected per per PriO unicast service and per PriO multicast service per port. When the TDM attribute is selected, it indicates that the corresponding packet carries the TDM service, otherwise, it indicates that the corresponding packet carries the packet service. Each port per Pril, Pri2, Pri3 unicast service and per-Pril, Pri2 Pri3 multicast service per port has no TDM attribute, and can only forward packet services. The USB is shared by all the port unicast services of the switching component of the packet switching device, and the dual-port memory structure separated by read and write operations is used to store the service data in units of packets, and the writing rate is not lower than that of the switching component. The sum of the maximum packet input rates of the input ports, and the read rate is not lower than the sum of the maximum packet output rates of all output ports of the switching unit. QM uses the threshold to control the USB usage of the unicast service. As shown in Figure 3, the USB High Threshold (USB_HT) is the total capacity of the USB. The USB Low Threshold (USB-LT) is the capacity that the USB has allocated to the switching unit for all unicast priority traffic queues and unicast TDM traffic queues. USBJLT must be smaller than USB-HT. The buffer capacity from USB-LT to USB-HT is reserved for use by TDM unicast services. When the USB usage reaches USB-LT, it is prohibited to continue to buffer unicast packets of any packet service type. As long as the USB usage does not reach USB-HT, it is always allowed to buffer any unicast TDM service packets.

The MSB is shared by all the ports of the switching component. The dual-port memory structure separated by read and write operations stores the service data in units of packets. The write rate is not lower than the sum of the maximum packet input rates of all input ports of the switching component. The output rate is not lower than the sum of the maximum packet output rates of all output ports of the switching unit.

The QM uses the threshold P threshold to control the occupancy of the MSB by the multicast service. As shown in Figure 4, the MSB High Threshold (MSB_HT) is the total capacity of the MSB. The MSB Low Threshold (MSB-LT) is the capacity that the MSB has allocated to the switching component for all multicast priority service queues and multicast TDM service queues. MSB—LT must be less than MSB—HT. The buffer capacity from MSB-LT to MSB-HT is reserved for use by TDM multicast services. When the occupancy of the MSB reaches MSB_LT, it is prohibited to continue to buffer multicast packets of any packet service type. As long as the MSB_HT is not occupied by the MSB, it is always allowed to buffer any multicast TDM service packets. The four unicast priority queues included in each port are Unicast Priority Queue 0 (UPQ0 for short), UPQ1, UPQ2, and UPQ3. These UPQs have the same structure and use FIFO operation, as shown in Figure 5. The UPQ unit stores the retrieval address of a particular unicast priority service packet, rather than the data contained in a particular priority unicast service packet. The unicast packet storage address is always output from the UPQ Head (UPQ_H), and is added from the next unit of the UPQ Tail (UPQ-T). QM uses the threshold to limit the USB usage per UPQ per port. As shown in Figure 6, each port contains UPQ Lower Threshold (UPQJLT) and UPQ High Threshold (UPQ High Threshold, UPQ-HT). ). UPQ—HT is the storage space for a particular UPQ. The UPQ-LT is used to control the occupation of the USB by the unicast service corresponding to the specific UPQ. When the packet queuing length of the specific UPQ is lower than the UPQ-LT, the USB is allowed to continue to buffer the corresponding unicast service packet. For each UPQ, the set UPQ-LT cannot be greater than UPQ-HT. UPQ of all UPQs - the sum of LT Equal to USB-LT equivalent unicast packet storage address number. Any unicast packets currently queued by the UPQ are not continuously stored in the USB. As shown in Figure 7, unicast packets belonging to the same UPQ can be distributed anywhere in the USB. The four multicast priority queues included in each port are Multicast Priority Queue 0 (MPQ0), MPQ1, MPQ2, and MPQ3. These MPQs have the same structure and use FIFO operation, as shown in Figure 8. The MPQ unit stores the retrieval address of a particular multicast priority service packet, rather than the data contained in a particular priority multicast service packet. The multicast packet storage address is always output from the MPQ Head (MPQ Head, MPQ-H for short), and is added from the next unit of the MPQ Tail (MPQ-T). QM uses the threshold to limit the MSB occupancy per MPQ per port. As shown in Figure 9, each MPQ includes MPQ Lower Threshold (MPQ-LT) and MPQ High Threshold (MPQ). — HT ). MPQ_HT is the storage space of a specific MPQ. The MPQJLT is used to control the occupancy of the MSB by the multicast service corresponding to the specific MPQ. When the packet queuing length of the specific MPQ is lower than the MPQ-LT, the MSB is allowed to continue to buffer the corresponding multicast service packet. For each MPQ, the set MPQ-LT cannot be greater than MPQ-HT. The sum of MPQ_LT of all MPQs is equal to the number of MSB-LT equivalent multicast packet storage addresses.

The QM uses a Queue Input Grant (QIG) signal to notify the outside world of incoming packets. Each UPQ and each MPQ has a corresponding QIG signal. For unicast packet services, if any UPQ queue length is greater than the UPQ_LT of the UPQ, the QM sets the QIG signal corresponding to the UPQ to true, otherwise it is set to false; for multicast packet services, any MPQ queue length Below the MPQ_LT of the MPQ, the QM sets the QIG signal corresponding to the MPQ to true, otherwise it is set to false. For the unicast TDM service, if the current occupation of the USB is lower than the USB-HT, the QM sets the QIG signal corresponding to the UPQ to be true, otherwise it is set to be a branch; for the multicast TDM service, if the current occupancy of the MSB is lower than When MSB-HT, QM sets the QIG signal corresponding to the MPQ to true, otherwise it is set to £. Any unicast packets currently queued by the MPQ are not continuously stored in the MSB. As shown in FIG. 10, multicast packets belonging to the same MPQ may be distributed anywhere in the MSB. As can be seen from Figure 10, for any multicast packet, the QM retains only one copy at the MSB, copying the storage address of the multicast packet to the corresponding multicast queue for multiple designated ports. For example, "multicast packet y" needs to be multicast to port 1. Port j, port k, and corresponding storage addresses are copied to MPQi, MPQj, and MPQk, respectively.

The PSch schedules service packets from UPQs or MPQs according to a fixed-length Scheduling Time Slot (STS). Each port contains a PSch, which is responsible for scheduling 4 UPQs and 4 MPQs for that port. The PSch of each port works independently. The total number of STSs that can be processed per second by all PSch is equivalent to the maximum number of packets that the switching component can forward per second, corresponding to the maximum packet switched capacity of the switching component. PSch is based on priority scheduling packets. Combined with Figure 2 and Figure 11, the scheduling process is described as follows: Although the STSs of specific ports are not contiguous, the time interval between neighboring STSs is the same, and is an integer multiple of STS, PSch is only in every STS. The scheduling operation can be performed during the period. Each STS can schedule one packet from the four UPQs and four MPQs of the corresponding port. The PSch of each port has eight Queue Output Grant (QOG) input signals, which are respectively controlled. Scheduling of 4 UPQs and 4 MPQs on a port; UPQ and MPQ of the same Pri on a specific port have the same scheduling priority, PSch can preferentially schedule UPQ, or preferentially schedule MPQ, or can be randomly selected; when a specific PSch performs scheduling First, the packet of the PriO queue is first scheduled. If the PriO queue is non-empty and the corresponding QOG signal status is true, the PSch outputs a packet from the corresponding PriO queue, and then enters the waiting state, waiting for the arrival of the next STS; otherwise If the PRL queue is not empty and the corresponding QOG signal status is true, PSch outputs a packet from the corresponding PRL queue, after which Entering the wait state, waiting for the arrival of the next STS; and so on, until the Pri3 queue is scheduled; if there is no suitable packet output during the current STS, the PSch goes directly to the wait state, waiting for the arrival of the next STS; Outputting a unicast packet from the selected UPQ, ₍ immediately releasing the storage address of the unicast packet; the specific PSch outputs a multicast packet from the selected MPQ each time, if the corresponding port is the last multicast port of the multicast packet , immediately release the storage address of the multicast packet, otherwise, continue to maintain the storage address of the multicast packet; when a specific PSch outputs a packet, simultaneously output a multicast output indication (Multicast Output Indicating, MCOI) The signal, when the MCIO signal is true, indicates that the packet multicast packet is currently being output from the MSB, otherwise, the unicast packet is output from the USB.

The AM is responsible for managing the free storage addresses of the USB and MSB. When a specific service needs to buffer a unicast or multicast input packet, AM allocates a USB or MSB free storage address for the queue corresponding to the service. Each PSch releases a unicast or multicast storage address, and the AM immediately reclaims the storage address. As described above, the packet switching method performed by the packet switching apparatus according to the embodiment of the present invention is as shown in FIG. 12, and includes the following steps: S1202: The grouping unit divides or fills the input data into fixed-length packets; and S1204, exchanges The unit will separately unicast packets and multicast packets from the packet unit Dividing into at least one priority and assigning a queue to each of the unicast packets and the multicast packets of each priority, and scheduling the packets according to the priority of the queue, wherein the queues assigned to the highest priority unicast packets are allocated The queue of the highest priority multicast packet is used for 7-bit TDM traffic or packet traffic, and the queues assigned to other priority unicast packets and the queues assigned to other priority multicast packets are used to carry packet traffic. The queue is used to buffer the address of the packet and operates in a first-in, first-out manner. Step S1204 includes: S12042, storing a unicast packet from a packet unit through a unicast shared buffer and using a unicast shared buffer space allocated for a highest priority unicast packet queue for TDM unicast traffic or unicast packet Traffic, and storing multicast packets from the packet unit through the multicast shared buffer and allocating the multicast shared buffer space for the highest priority multicast packet queue for TDM multicast traffic or multicast packet traffic; S12044. A packet scheduling unit, configured to schedule a packet from a queue according to a priority of the queue; and S12046, after the unicast packet and the multicast packet are forwarded to the designated port, the shared by the dry unicast packet and the multicast packet Buffer space. Wherein, the write speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet input rates of all the input ports of the switching unit, and the reading of the unicast shared buffer and the multicast shared buffer The output speed is not lower than the sum of the packet output speeds of all the output ports of the switching unit. The multicast shared buffer copies the storage address of the multicast packet to the corresponding multicast queue of the plurality of designated ports, and releases the multicast packet storage address when the predetermined time is reached. Wherein, in the case that the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the occupied multicast shared buffer space is lower than In the case of assigning a value, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. Wherein, in the case that the unicast shared buffer has free storage space, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and the idle storage space exists in the multicast shared buffer. Next, it is allowed to buffer the currently input multicast packet, otherwise the currently input multicast packet is discarded. The unicast service and the multicast service of the same priority are scheduled in the following manner: preferentially scheduling the unicast service, preferentially scheduling the multicast service, and scheduling the random. In summary, the core idea of the present invention is: to reserve a certain global buffer space for the highest 4 priority service, to buffer the TDM service based on the global shared buffer threshold, and to buffer the packet service based on the traffic queue threshold. Grouping. The port service scheduling is performed according to the priority, and the highest priority service is preferentially forwarded. Use separate unicast shared buffers and multicast shared buffers, and set up separate unicast traffic queues and multicast traffic queues for each port to avoid interaction between the two. Specifically, the process of implementing the core idea of the present invention mainly includes the following steps: Step 1: The packet component of the packet switching device according to the present invention exchanges services according to fixed length packets, wherein any service data accessed from the user interface must be segmented. Or a fixed-length packet of a specific length can be sent to the switching component of the packet switching device. The switching component in the packet switching apparatus according to the present invention divides the unicast service into one or more priority levels, and allocates one queue for each priority per-port single "service; the multicast service is also divided into one or more priorities. Level, and allocate a queue for each priority multicast service per port. Among them, the queue is used to queue the storage address of the buffer packet, and operates according to the First Input First Output (FIFO) mode. Broadcast shared buffer, multicast shared buffer, packet scheduling, queue management, address management, etc. Step two, unicast shared buffer stores unicast packets; multicast shared buffer stores multicast packets Both buffers are dual-port storage structures, one for writing packets and the other for reading packets, prohibiting simultaneous reading and writing of the same storage address. Shared unicast buffer and shared The write rate of the broadcast buffer is not lower than the sum of the packet input rates of all the input ports of the switching device, and the read speed is not lower than the packet input of all the output ports of the switching device. The sum of the rates ensures that there is no input and output congestion. Step 3: The queue management component allocates a specific unicast shared buffer space for each port per priority unicast queue, and reserves part of the unicast shared buffer space for TDM. Unicast service; allocates a specific multicast shared buffer space for each port per priority multicast queue, and reserves part of the multicast shared buffer space for TDM multicast services. Step 4: For any multicast packet, queue The management component only keeps one copy in the multicast shared buffer, and copies the storage address of the multicast packet to the corresponding multicast queue of multiple designated ports. The queue management component sets a counter for any multicast packet to control multicast. The release of the packet storage address. Only after the multicast packet has been forwarded to all designated ports, the queue management component can release the storage space corresponding to the multicast packet. Step 5, all the highest priority unicast and multicast Grouping has optional TDM attributes, if The current packet contains a TDM attribute, indicating that the current packet carries the TDM service, and otherwise indicates that the current packet carries the packet service. Step 6: For the unicast packet service, if the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise, the currently input unicast packet is discarded. For multicast packet services, if the occupied multicast shared buffer space is lower than the assigned value, it is allowed to buffer the currently input multicast packet, otherwise, the currently input multicast packet is discarded. Step 7: For the unicast TDM service, if the unicast shared buffer has free storage space, it is allowed to buffer the currently input unicast packet, otherwise, the currently input unicast packet is discarded. For multicast TDM services, if the multicast shared buffer has free storage space, it is allowed to buffer the currently input TDM multicast packet, otherwise, the currently input TDM multicast packet is discarded. Step 8: The packet scheduling component schedules the service according to the priority, first schedules the highest priority service, and finally schedules the lowest priority service. The unicast service and the multicast service of the same priority belong to the same scheduling level, and the unicast service or the multicast service may be preferentially scheduled, or may be randomly selected. In step nine, the address management component manages the free storage addresses of the unicast shared buffer and the multicast shared buffer. One packet can be buffered per storage address. The present invention uses a separate unicast shared buffer and a multicast shared buffer, and sets a separate queue for each unicast priority service per port, and per port per multicast priority service, when the switching component forwards In the case of a connectionless packet service, such as an Ethernet Media Access Control (MAC) frame, a MAC frame of an unknown destination address can be converted into a fixed-length multicast packet forwarding, thereby avoiding the address learning process. The impact of broadcasting business. In addition, the present invention buffers the packet service according to the specific queue threshold, buffers the TDM service according to the unicast shared buffer threshold and the multicast shared buffer threshold, and preferentially schedules the TDM service, and ensures the effectiveness of the TDM service forwarding. At the same time, it can ensure that the forwarding of TDM services is not affected by packet traffic congestion. The above is only the preferred embodiment of the present invention, and is not intended to limit the present invention, and various modifications and changes can be made to the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and scope of the present invention are intended to be included within the scope of the present invention.

Claims

Claims: A packet switching device, characterized by including:

a grouping unit, used to divide or fill the input data into fixed-length packets; and a switching unit, used to divide the unicast packets and multicast packets from the grouping unit into at least one priority and provide each priority with The unicast packets and multicast packets are allocated to a queue respectively, and are scheduled according to the priority of the queue. Among them, the queue allocated to the highest priority unicast packet and the queue allocated to the highest priority multicast packet are used to carry TDM services or packet services, queues allocated to unicast packets of other priorities and queues allocated to multicast packets of other priorities are used to carry the packet services. The packet switching device according to claim 1, wherein the queue is used to cache the address of the packet and operates in a first-in, first-out manner. The packet switching device according to claim 1, wherein the switching unit includes: a unicast shared buffer for storing unicast packets from the packet unit; a multicast shared buffer for storing multicast packets from the grouping unit; a packet scheduling unit configured to schedule the packets from the queue according to the priority of the queue;

The queue management unit is used to allocate shared buffer space to the unicast packet queue and multicast packet queue of each port, and use the unicast shared buffer space allocated for the highest priority unicast packet queue for TDM single broadcast service or unicast packet service, the multicast shared buffer space allocated to the highest priority multicast packet queue is used for TDM multicast service or multicast packet service, and between the unicast packet and the multicast After the packet is forwarded to the designated port, release the shared buffer space corresponding to the unicast packet and the multicast packet; and

An address management unit, configured to manage the free storage addresses of the unicast shared buffer and the multicast shared buffer period. The packet switching device according to claim 3, characterized in that, the writing speed of the unicast shared buffer and the multi-" shared buffer is not lower than the packet input of all input ports of the switching unit The sum of the rates, the read speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet output speeds of all output ports of the switching unit. According to claim 3 Packet switching device, characterized in that the queue management unit copies the storage address of the multicast packet in the multicast shared buffer to the corresponding multiple queues of multiple designated ports, and releases it when the predetermined time is reached. The multicast packet storage address. The packet switching device according to claim 4, characterized in that:

When the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and

When the occupied multicast shared buffer space is lower than the allocated value, the currently input multicast packet is allowed to be buffered, otherwise the currently input multicast packet is discarded. The packet switching device according to claim 4, characterized in that:

When there is free storage space in the unicast shared buffer, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and

If there is free storage space in the multicast shared buffer, the currently input multicast packet is allowed to be buffered; otherwise, the currently input multicast packet is discarded. The packet switching device according to claim 3, characterized in that the queue scheduling unit schedules unicast services and multicast services of the same priority in the following manner: unicast services are prioritized, multicast services are prioritized, and multicast services are scheduled randomly. . The packet switching device according to claim 3, wherein the queue scheduling unit prioritizes packets in the highest priority queue. A packet switching method, characterized by including:

S 1202, the grouping unit divides or fills the input data into fixed-length groups; and

S1204. The switching unit divides the unicast packets and multicast packets from the packet unit into at least one priority and allocates the unicast packets and multicast packets to each priority level respectively. A queue, and the packet is scheduled according to the priority of the queue, wherein the queue allocated to the highest priority unicast packet and the queue allocated to the highest priority multicast packet are used to carry TDM services or packet services , the queues allocated to unicast packets of other priorities and the queues allocated to multicast packets of other priorities are used to carry the packet service.

11. The packet switching method according to claim 10, wherein the queue is used to cache the address of the packet and operates in a first-in, first-out manner.

12. The packet switching method of claim 10, wherein step S1204 includes:

S 12042. Store unicast packets from the packet unit through a unicast shared buffer and use the unicast shared buffer space allocated for the highest priority unicast packet queue for TDM unicast services or unicast packet services. , and storing multicast packets from the packet unit through a multicast shared buffer and using the multicast shared buffer space allocated for the highest priority multicast packet queue for TDM multicast service or multicast packet service; as well as

S 12044, a packet scheduling unit, configured to schedule the packet from the queue according to the priority of the queue; and

S 12046: After the unicast group and the multicast group are forwarded to the designated port, release the shared buffer space corresponding to the unicast group and the multicast group.

13. The packet switching method according to claim 12, characterized in that the writing speed of the unicast shared buffer and the multicast shared buffer is not lower than the packet speed of all input ports of the switching unit. The sum of the input rates, the read speed of the unicast shared buffer and the multicast shared buffer is not lower than the sum of the packet output speeds of all output ports of the switching unit.

14. The packet switching method according to claim 12, characterized in that, the multicast shared buffer copies the storage address of the multicast packet to the corresponding multicast queues of multiple designated ports, and when the predetermined time is reached, Release the multicast packet storage address.

15. The packet switching method according to claim 12, characterized in that:

When the occupied unicast shared buffer space is lower than the allocated value, the currently input unicast packet is allowed to be buffered, otherwise the currently input unicast packet is discarded; and When the occupied multicast shared buffer space is lower than the allocated value, the currently input multicast packet is allowed to be buffered; otherwise, the currently input multicast packet is discarded.

16. The packet switching method according to claim 12, characterized in that:

If there is free storage space in the multicast shared buffer, the currently input multicast packet is allowed to be buffered; otherwise, the currently input multicast packet is discarded.

17. The packet switching method according to claim 12, characterized in that unicast services and multicast services of the same priority are scheduled in the following manner: unicast services are prioritized, multicast services are prioritized, and multicast services are scheduled randomly.