WO2013053276A1 - 一种网络路由收敛处理方法和装置 - Google Patents
一种网络路由收敛处理方法和装置 Download PDFInfo
- Publication number
- WO2013053276A1 WO2013053276A1 PCT/CN2012/081571 CN2012081571W WO2013053276A1 WO 2013053276 A1 WO2013053276 A1 WO 2013053276A1 CN 2012081571 W CN2012081571 W CN 2012081571W WO 2013053276 A1 WO2013053276 A1 WO 2013053276A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- node
- network
- fast channel
- lsp
- packet
- Prior art date
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/02—Topology update or discovery
- H04L45/03—Topology update or discovery by updating link state protocols
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/22—Alternate routing
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L45/00—Routing or path finding of packets in data switching networks
- H04L45/50—Routing or path finding of packets in data switching networks using label swapping, e.g. multi-protocol label switch [MPLS]
Definitions
- the present invention relates to a communication network, and in particular, to a network route convergence processing method and apparatus.
- SPB Shortest Path Bridging
- TRILL Transparent Interconnection of lots of Links
- ISIS Intermediate System to Intermediate System
- SPB technology and TRILL technology have advantages over traditional Layer 2 technologies, such as Equal-Cost Multipath (ECMP), high bandwidth utilization, and suitable for specific data center networking.
- ECMP Equal-Cost Multipath
- the convergence time of the entire network is an important performance indicator.
- the information diffusion of the network topology changes is layer by layer in a wave manner. For example, if a network includes 7 node devices, the node device 1 fails, and the nodes 2, 3, 4, and 5 at the same network layer detect the fault, and notify the CPU of the respective device after detecting the fault (Central Processing Unit, The central processing unit performs processing on the control plane. After the processing at the CPU control level is completed, a Link State Protocol Data Unit (LSP) packet is generated, and the generated LSP packet is sent to the outside through the normal channel.
- LSP Link State Protocol Data Unit
- the neighboring nodes 6 and 7 of the layer, and the CPU control planes of the nodes 6 and 7 can implement network convergence only after the LSP packet processing is completed.
- the diffusion rate of the topology change information is relatively slow.
- the network convergence time requires at least the CPU processing time of the node 3, the protocol 4 transmission time t2+, and the CPU processing time t3 of the node 6, It is tl+t2+t3, and the larger the network topology, the worse the convergence performance.
- the convergence performance of the network is related to the location of the topology change. If the topology change occurs at the edge of the network, the synchronization of the LSPs of the entire network needs to span the entire network diameter, even if the topology changes. Born in the center of the network, the synchronization of the entire network of LSP 4 ⁇ text also needs to traverse the network radius. Summary of the invention
- the present invention provides a network route convergence processing method for solving the above problems, and the method includes:
- a network route convergence processing device configured to pre-establish a fast channel for connecting devices in the network, where the fast channel is a prune-free multicast distribution tree;
- topology change awareness module for detecting topology change events
- a protocol packet generating module configured to generate a link state protocol data unit LSP packet according to the type of the topology change event
- the protocol packet sending module is configured to send the LSP packet according to the multicast routing table corresponding to the fast channel.
- the implementation of the embodiment of the present invention has the following beneficial effects: the pre-established prune-free multicast distribution tree is used as a fast channel to spread the topology change information, because the convergence processing start time of each device on the fast channel is small, Basically in the state of parallel processing, the convergence performance of the entire network is greatly improved, and the convergence time of the entire network is basically independent of the network scale and topology change position.
- FIG. 2 illustrates a first topological variation of a network in accordance with an embodiment of the present invention.
- FIG. 3 illustrates a schematic diagram of a topology change of a network according to an embodiment of the present invention.
- FIG. 4 illustrates a second topological variation of a network in accordance with an embodiment of the present invention.
- Fig. 5 is a block diagram showing the structure of a network route convergence processing apparatus according to an embodiment of the present invention.
- a fast channel for connecting devices in the network is established in advance, where the fast channel is a prune-free multicast distribution tree;
- the neighboring devices of the topology change detect a topology change event, and generate a link state protocol data unit LSP packet according to the type of the topology change event, and prepare the root device.
- the network in the embodiment of the present invention can run the multi-link transparent interconnection TRILL protocol, and the network running the TRILL protocol is referred to as a TRILL network.
- the network in the embodiment of the present invention can also run the shortest path bridging SPB protocol, and the network running the SPB protocol is referred to as an SPB network.
- the route convergence processing method provided by the present invention can be applied to both the TRILL network and the SPB network.
- the route convergence processing method provided by the present invention can also be applied to other networks than the above two networks that can construct a pruned fast channel.
- the topology change in the embodiment of the present invention includes a link state change and a node device state change, where the link state change includes a link fault, and the node device state change includes a node device fault and a node setup overload.
- Topological change When a link is faulty, each neighboring device that changes topology is a device that is directly connected to the failed link. If the faulty link is directly connected, there are two devices, such as the network shown in Figure 2.
- the node directly connected to the failed link has a node 6 and a node 8, and then the neighboring devices of the topology change referred to in the present invention refer to the node 6 and the node 8.
- the faulty link is directly connected to three devices, for example, in the network architecture shown in FIG. 3, the node directly connected to the failed link has node 1, node 2, and node 3, then the extension referred to in the present invention
- Each adjacent device of the Park change refers to Node 1, Node 2, and Node 3.
- each adjacent device of the topology change refers to each device directly connected to the node device.
- the node 6 fails, and The node device directly connected to the node 6 is the node 3 and the node 8.
- the neighboring devices of the topology change referred to in the present invention are the node 3 and the node 8. If the node 8 fails, the node device directly connected to the node 8 is a node. 5. Node 6 and Node 7, then the neighboring devices of the topology change referred to in the present invention are Node 5, Node 6, and Node 7.
- the neighboring devices of the topology change generate LSPs according to the detected topology changes, and the generated LSPs are forwarded through the fast channel to notify other devices of the topology change event, and then continue Route convergence processing is performed for the topology change event.
- the fast channel in S100 is a prune-free multicast distribution tree.
- Different networks can establish a prune-free multicast distribution tree as a fast channel for topology change information diffusion according to the specific provisions of its corresponding protocol.
- each device in the network itself contains the multicast routing table corresponding to the fast channel it is in.
- the device forwards the packet to the multicast routing table.
- the device corresponding to the next routing entry in the multicast routing table. If the next routing entry in the multicast routing table is empty, the node is the end node on the fast channel.
- the LSP does not need to be forwarded. .
- the underlying chip can be used to implement the multicast routing table search, and the LSP packet can be sent to the device on the fast channel according to the multicast routing table corresponding to the fast channel, and the node 1 is taken as an example.
- the device on the fast path of the node 1 is the corresponding device in the multicast routing table corresponding to the fast channel of the node 1. If the multicast forwarding table corresponding to the fast channel where the node 1 is located is empty, the node 1 is the fast. The last node on the channel, the LSPs sent to the node do not need to continue to be forwarded through the fast channel.
- the generated LSP message is forwarded to the device on the fast channel according to the multicast routing table corresponding to the fast channel, where the device on the fast channel includes The topology changes the corresponding devices in the respective multicast routing tables of the neighboring devices. If one of the neighboring devices in the topology change is the last node device on the fast path, the neighboring device can query the multicast routing table of the device to learn that the generated LSP packet cannot be forwarded through the fast channel.
- the network route convergence processing method provided by the present invention further includes: the device on the fast channel receives the neighboring device of the topology change through the fast channel And sending the LSP packet to the device corresponding to the multicast routing table according to the multicast routing table corresponding to the fast channel where the device is located on the fast channel, and sending the LSP packet
- the convergence processing module of the device sent to the fast channel performs route convergence processing.
- the specific convergence processing of the convergence processing module includes: recalculating the unicast and multicast routing tables, and sending the calculated unicast and multicast routing table.
- the protocol packet sending module can use the underlying chip to forward LSP packets, and the underlying chip finds the multicast routing table corresponding to the fast channel. Then, the LSP packet is copied and sent to the egress port of the device corresponding to the multicast routing table. Because the underlying forwarding efficiency is very high, the forwarding time is generally much smaller than the route convergence processing time, so that the route convergence processing of each device on the fast channel is basically in a parallel processing state.
- the steps of the fast track establishment may include:
- the neighbor establishment and database synchronization are performed according to the ISIS protocol.
- the device that runs the TRILL protocol with the highest root priority is the first to specify the root of the tree.
- the device running the TRILL protocol specifies the maximum number of multicast trees to be established. K and S multicast tree roots, where
- the roots of the S multicast tree are used as the root of the fast channel.
- the devices running the TRILL protocol in the network are sorted according to the root priority, and the first K devices are used.
- the S multicast tree roots are used as the first S roots of the fast channel, and devices other than the S multicast tree roots in the network are prioritized by the root.
- the sorted first KS devices are used as the KS roots of the fast channel.
- the device running the TRILL protocol completes neighbor establishment and database synchronization according to the ISIS protocol, and finally each TRILL protocol in the network runs.
- Devices have topology information for the entire network.
- Each device running the TRILL protocol can advertise the maximum number of multicast trees it can calculate through LSP packets.
- Tree Root Priority The highest device that runs the TRILL protocol can preferentially specify the root of the tree.
- the device running the TRILL protocol specifies the maximum number of multicast trees that can be established in the network, K and S multicast tree trees (eg, nickname 1, nickname2, nickname3... nickname S).
- the S multicast tree roots specified by the device running the TRILL protocol are used as the root of the fast channel; when S is equal to zero, the devices running the TRILL protocol in the network are sorted according to the root priority.
- the first K are the roots of the fast channel; when S is less than K, the S multicast tree roots specified by the device running the TRILL protocol are used as the first S roots of the fast channel, and the specified tree is excluded from the network.
- the devices other than the root are sorted according to the root priority, and the sorted first KS devices are used as the KS roots of the fast channel.
- the maximum number of multicast trees specified by the device is 4, and the devices other than Tx and Ty in the network are sorted according to the root priority.
- the other two roots if the network also includes devices Ta, Tb, and Tc, the order of tree root priority in the network is Ty>Ta>Tc>Tb>Tx, then the root of the K tree is ⁇ Tx, Ty , Ta, Tc ⁇ .
- Each device is based on the tree root independent calculation of the pruned multicast distribution tree obtained by the above calculation.
- all devices are connected at the forwarding level, which is equivalent to a virtual ether.
- the interface connects each device in the network to each other.
- any node device in the network can serve as the root of the fast channel.
- the prune-free multicast distribution tree can be established as a fast channel for topology change information diffusion according to the corresponding protocol specification.
- the device in the SPB network or the TRILL network needs to perform neighbor check and RPF (Reverse Path Forwarding) check to prevent loops.
- RPF Reverse Path Forwarding
- topology changes of the network may include link state changes, such as links Up/down, cost change, etc., can also include node device status changes, such as node device failure, setting overload, etc.
- Setting overload refers to the network administrator needs to upgrade or maintain a node. The node is set to the overload state. The node notifies the entire network that the node other than the node has overload information through the LSP packet, so that other nodes bypass the node when calculating the routing table.
- FIG. 2 a schematic diagram of a first topology change of a network according to an embodiment of the present invention is illustrated.
- the network includes eight node devices, and node 8 is a tree root. As shown in Fig. 2, the link between the node device 6 and the node device 8 changes.
- the root of the pre-established fast channel is node 8
- each branch with node 8 as the root of the tree is ⁇ node 5, node 2 ⁇ , ⁇ node 6, node 3, node 1 ⁇ , ⁇ node 7, node 4 ⁇ .
- the specific steps of processing the network route convergence processing method of the present invention may be performed by using the pre-established fast channel.
- Node 6 and node 8 generate an LSP packet according to the type of the topology change event
- the node 6 and the node 8 send the generated LSP message to the device on the fast channel through the multicast routing table corresponding to the fast channel, and then the node 6 and the node 8 continue to perform route convergence processing on the topology change event, specifically , including recalculating the unicast and multicast routing tables, and sending the calculated unicast and multicast routing tables to the protocol packet sending module;
- the device on the fast path of the node 6 and the node 8 receives the LSP packet sent by the fast channel, and forwards the LSP 4 to the device corresponding to the multicast routing table through the multicast routing table corresponding to the fast channel, and The received LSP message is sent to the convergence processing module of the device for performing convergence processing.
- the method includes: recalculating the unicast and multicast routing table, and sending the calculated routing table to the protocol packet sending module for reporting. Forwarding, where,
- the devices on the fast path of the node 6 and the node 8 and the node 6 and the node 8 send the LSP message to the neighbor node of the device itself through the normal channel after the device itself completes the convergence process.
- the neighboring device node 6 and the node 8 whose link state changes detect a topology change event, that is, a link state change, according to a type of topology change event (for example, a chain
- the LSPs generated by the LSPs are sent to the fast multicast routing table.
- the device on the speed channel, for node 6, the node device corresponding to the next entry in the multicast routing table corresponding to the fast channel is node 3, and the multicast routing table corresponding to the fast channel where node 3 is located
- the node device corresponding to the next entry is node 1.
- a topology change event that is, a link state change
- a type of topology change event for example, a chain
- the LSPs generated by the LSPs are sent to the fast multicast routing table.
- the device on the speed channel, for node 6, the node device corresponding to the next entry in the multicast routing table corresponding to the fast channel is node 3, and the multicast routing table corresponding to the fast channel where node
- the next entry of the multicast routing table corresponding to the fast channel where the node 1 is located is empty, and the node 1 receives the fast After the LSP is sent, the LSP is forwarded through the fast channel.
- the next entry in the multicast routing table is empty.
- the devices on the fast channel are node 7 and node 4, and node 5 and node 2. The process of forwarding the LSP through the fast channel is similar, and will not be described here.
- the specific step of forwarding includes: searching, by the underlying chip, the multicast routing table corresponding to the fast channel where the device itself is located, and copying and transmitting to the egress port of the corresponding device in the multicast routing table according to the found multicast routing table;
- the node 6 and the node 8 continue to perform route convergence processing on the topology change event.
- the specific steps of the convergence process include newly calculating the unicast and multicast routing table, and sending the calculated unicast and multicast routing table to the protocol report.
- the text transmission module is configured to forward the packet.
- the LSP packet that is flooded in the fast channel is an LSP packet generated according to the control protocol itself, so the message digest algorithm of the protocol itself can be used for security processing, thereby avoiding network attacks and ensuring the network.
- the nodes corresponding to the devices on the fast channel of node 6 and node 8 respectively 3 and node 1, node 7, node 4, node 5, and node 2, because the processing of each node device is similar, the following takes node 3 as an example to describe the processing procedure of the device to forward the LSP packet forwarded on the fast channel.
- the node 3 After receiving the LSP packet sent by the fast channel, the node 3 searches for the multicast routing table corresponding to the fast channel through the underlying chip, and sends the received LSP packet to the multicast routing table according to the found multicast routing table.
- the egress port of the device corresponding to the multicast routing table is copied and sent.
- the node 3 After receiving the LSP packet sent by the fast channel, the node 3 sends the received LSP packet to the convergence of the node 3.
- the processing module performs convergence processing, specifically, including calculating a unicast and multicast routing table, and sending the calculated unicast and multicast routing table to the protocol packet sending module for packet forwarding, where the recalculated multicast
- the routing table can be used to update the multicast routing table corresponding to the original fast channel. All the devices in the network receive the LSPs with the topology change and receive the LSP according to the received LSP. After the convergence process is completed, the entire network is converged.
- the device in the network forwards the LSP packet through the fast channel, and the general transmission time is in the microsecond level, usually small and small.
- the convergence time is several hundred milliseconds with respect to the existing network, and the time is basically negligible.
- the start time of each device convergence processing differs only at the microsecond level, and is basically in parallel processing.
- the state of the network makes the convergence time of the entire network substantially equal to the convergence time of a single node device, and the convergence of the entire network is basically no longer limited by the network size and topology change position, which greatly improves the convergence performance of the network.
- the devices on the fast path of the node 6 and the node 8 and the node 6 and the node 8 can forward the LSP message to the neighbor node connected thereto through the normal channel after the device self-convergence processing is completed.
- the normal channel described in this document refers to the message transmission channel already existing between devices in the network. In this channel, LSP packets are serially transmitted, which is called "positive channel".
- the devices in the network may receive duplicate LSPs from the normal channel and the fast channel.
- the network's respective protocols can automatically handle the repeated transmission. In the case of repeated transmission, the repeated reception is repeated. The message is not processed.
- the new multicast distribution tree is recalculated to provide a new fast path for subsequent new topology changes.
- FIG. 4 a schematic diagram of a second topology change of a network according to an embodiment of the present invention is illustrated.
- the network includes eight node devices, and node 8 is the root of the tree.
- the state of the node device 6 changes, which may be a node failure or an overload setting.
- the root of the pre-established fast channel is node 8
- each branch with node 8 as the root is ⁇ node 5, node 2 ⁇ , ⁇ node 6, node 3, node 1 ⁇ , ⁇ node 7, node 4 ⁇ .
- the state of the node 6 changes.
- the link changes involved include the state changes of the two links between node 3 and node 6 and between node 8 and node 6.
- the specific steps of processing the network route convergence processing method of the present invention may be performed by using the foregoing fast path.
- Adjacent device nodes 3 and 8 detecting node state changes detect topology change events
- Node 3 and node 8 generate an LSP packet according to the type of the topology change event
- Node 3 and node 8 send the generated LSP packet through the multicast routing table corresponding to the fast channel. To the device on the fast channel, then node 3 and node 8 continue to perform route convergence processing for the topology change event, specifically, including recalculating the unicast and multicast routing tables, and calculating the calculated unicast and The multicast routing table is sent to the protocol packet sending module.
- the device on the fast channel receives the LSP message sent by the fast channel, and forwards the LSP message to the device corresponding to the multicast routing table through the multicast routing table corresponding to the fast channel, and the receiving The LSP packet is sent to the convergence processing module of the device for convergence processing.
- the method includes: recalculating the unicast and multicast routing table, and sending the calculated routing table to the protocol packet sending module to perform packet forwarding.
- the device on the fast channel sends the LSP packet to the neighbor node of the device through the normal channel.
- the neighboring node device of the device on the fast channel receives the devices on the fast channel where the LSP report forwarded by the normal channel is forwarded.
- the node 3 and the node 8 detect a state change event of the link associated with the node 6.
- the link state change detected by the node 3 is represented by event A3.
- the link state change detected by the node 8 is represented by event A8.
- Node 3 and node 8 respectively detect link state change events A3 and A8, and generate corresponding LSPs according to the type of link state change.
- node 3 and node 8 generate the generated LSP ⁇ through the fast channel.
- the multicast routing table is sent to the device on the fast channel.
- the underlying chip searches for the multicast routing table corresponding to the fast channel where the device is located, and sends the received LSP packet to the multicast routing table according to the found multicast routing table.
- the multicast routing table is copied and sent by the egress port of the device.
- the node 3 and the node 8 continue to perform route convergence processing on the topology change event, specifically, including recalculating the unicast and multicast routing tables, and
- the unicast and multicast routing table is sent to the protocol packet sending module, where the recalculated multicast routing table can be used to update the multicast routing table corresponding to the original fast channel.
- the device on the fast channel of node 3 is node 1
- the devices on the fast channel of node 8 are node 5 and node 2, and node 7 and node 4.
- the node 1 receives only the LSP packet corresponding to the event A3.
- the node 5 and the node 2, the node 7 and the node 4 only receive the LSP packet corresponding to the event A8, only the event A3.
- the nodes in the network can correctly calculate Calculate the forwarding routing table to complete the convergence process.
- the LSP message received through the fast channel needs to be forwarded to the device through the normal channel. Its own neighbor node device forwards. For example, for node 1, after receiving the LSP packet corresponding to the A3 event through the fast channel (Node 3 -> Node 1), the convergence process is performed, and the received LSP packet needs to pass through the normal channel (node 1-> Node 2, Node 1 -> Node 4) are sent to Node 2 and Node 4 to ensure that each node in the network can receive both event A3 and event A8, thereby correctly calculating the route forwarding path completion convergence process.
- the node 2 and the node 4 After receiving the LSP packet sent through the normal channel, the node 2 and the node 4 determine whether the sequence number of the received LSP packet is greater than the sequence number of the packet in the current database. If the packet is greater than, the LSP is reported. The text is the latest LSP packet and needs to be sent through the fast channel on which it resides. Through the above re-forwarding, it can be ensured that both the event A3 and the event A8 can be received by the nodes in the network, thereby correctly calculating the new route forwarding path, and providing a new fast channel for the subsequent new topology change. BRIEF DESCRIPTION OF THE DRAWINGS The network route convergence processing apparatus provided by the present invention will be specifically described below with reference to the accompanying drawings and embodiments.
- the network route convergence processing device 500 can include:
- the pre-processing module 502 is configured to pre-establish a fast channel that connects devices in the network, where the fast channel is a non-pruned multicast distribution tree;
- a topology change awareness module 504 configured to detect a topology change event
- the protocol packet generation module 506 is configured to generate a link state protocol data unit LSP packet according to the type of the topology change event.
- the protocol packet sending module 508 is configured to send the LSP packet according to the multicast routing table corresponding to the fast channel.
- the network route convergence processing device 500 may include:
- the convergence processing module 510 is configured to perform route convergence processing.
- the protocol packet receiving module 512 is configured to receive an LSP packet.
- the search processing module 514 is configured to search the multicast routing table for the received LSP message, and is further configured to send the received LSP message to the convergence processing module for convergence processing.
- the topology change sensing module, the protocol packet generating module, the protocol packet sending module, the protocol packet receiving module, the convergence processing module, and the search processing module of the network route convergence processing device may be configured in each network.
- each device in the network is in a peer-to-peer position, and the same functional module can be configured on each device.
- the same functional module can be configured on each device.
- only some of the functional modules of some devices may be used.
- the lookup processing module and the protocol message sending module may be implemented using an underlying chip, wherein the underlying chip may employ any chip known to those skilled in the art that is capable of implementing the functions described herein.
- the network in the embodiment of the present invention can run the multi-link transparent interconnection TRILL protocol.
- the network running the TRILL protocol is called a TRILL network.
- the network in the embodiment of the present invention may also run the shortest path bridging SPB protocol.
- the network running the SPB protocol is referred to as an SPB network.
- the route convergence processing device provided by the present invention can be applied to both the TRILL network and the SPB network. It should be noted that the route convergence processing apparatus provided by the present invention can also be applied to other networks than the above two networks capable of constructing a pruned fast path.
- the fast channel in the embodiment of the present invention is a pruning-free multicast distribution tree.
- Different networks may establish a prune-free multicast distribution tree as a fast channel for topology change information spreading according to the specific provisions of the corresponding protocols.
- the fast channel is equivalent to a virtual Ethernet interface that connects devices in the network to each other.
- devices in the SPB network or TRILL network need to perform neighbor check and RPF check on the device to prevent loops.
- the LSP When an LSP is sent through the fast-channel, the LSP is encapsulated in the inner layer of the TRILL protocol packet or the SPB protocol packet.
- the pre-processing module 402 in the network route convergence processing apparatus in the embodiment of the present invention may be specifically configured to complete the neighbor establishment and database synchronization according to the intermediate system to the intermediate system ISIS protocol after the TRILL network deployment is completed, and run the TRILL protocol.
- the device that runs the TRILL protocol with the highest root priority is the root of the tree.
- the device that runs the TRILL protocol specifies the maximum number of multicast trees and the roots of the multicast tree.
- the S multicast tree roots serve as the root of the fast channel.
- the devices running the TRILL protocol in the network are sorted according to the root priority, and the first K devices are used as the root of the fast channel.
- the S multicast tree roots are used as the first S roots of the fast channel, and devices other than the S multicast tree roots in the network are prioritized by the root.
- the sorted first KS devices are used as the KS roots of the fast channel.
- each device running the TRILL protocol in the network has the topology information of the entire network.
- Each device running the TRILL protocol can advertise the maximum number of multicast trees it can calculate through LSP packets.
- the device running the TRILL protocol with the highest root priority in the device running the TRILL protocol can preferentially specify the root of the tree.
- the first device running the TRILL protocol specifies the maximum number of multicast trees that can be established in the network K and S multicast tree roots (for example, nickname 1, nickname2, nickname3... nickname S).
- the S multicast tree roots specified by the device running the TRILL protocol are used as the root of the fast channel; when S is equal to zero, the devices running the TRILL protocol in the network are sorted according to the root priority.
- the first K are the roots of the fast channel; when S is less than K, the S multicast tree roots specified by the device running the TRILL protocol are used as the first S roots of the fast channel, and the specified tree is excluded from the network.
- the devices outside the root are sorted according to the root priority, and the sorted first KS devices are used as the KS roots of the fast channel.
- Each device is based on the tree root independent computing multicast distribution tree obtained through the above calculation, and all devices are connected at the forwarding level through the multicast routing table stored on each device.
- any node device in the network can serve as the root of the fast channel.
- the corresponding pre-processing module can establish a pruning-free multicast distribution tree as a fast channel for topology change information diffusion according to the specific provisions of the protocol.
- the topology change of the network may include link state change, such as link up/down, cost value change, etc., and may also include node device state changes, such as node device failure, setting overload, etc., setting overload If the network administrator needs to upgrade or maintain a node, set the node to an overload state. The node will notify the entire network other than the node that the node has overload information through the LSP packet. The node bypasses the node when it calculates the routing table.
- the topology change sensing module on the adjacent device node 6 and the node 8 of the link state change detects the topology change event, and the protocol packet generation module changes the type according to the topology.
- the LSP packet is forwarded according to the multicast routing table that is located on the device.
- the device's own convergence processing module continues to perform route convergence processing for the topology change event, specifically, including recalculating the unicast and multicast routing tables, and sending the calculated unicast and multicast routing table to
- the protocol packet sending module is configured to forward the packet, and the calculated multicast routing table can update the multicast routing table corresponding to the original fast channel.
- the protocol packet receiving module of the node 3 receives the LSP packet forwarded by the protocol packet sending module of the node 6 through the fast channel, and after receiving the LSP packet, the node 3 searches for the multicast routing table of the fast channel where the device itself is located. The LSP packet is forwarded according to the found multicast routing table. On the other hand, the convergence processing module performs route convergence processing according to the generated LSP packet. The processing for Node 5 and Node 7 is similar and will not be described here. The node 3 forwards the received LSP message to the node 1 , and the node 5 and the node 7 continue to forward the received LSP message to the node 2 and the node 4 respectively.
- the fast channel is used to forward the LSP packet
- the processing time is generally in the microsecond level.
- the time difference between the time that each device receives the LSP packet is microsecond.
- the processing of LSPs is basically parallel, so that the convergence performance of the entire network is close to that of a single device, and the convergence of the entire network is basically independent of the network size and The limits of the topological change position.
- the protocol packet sending module of the device may be used to send the received LSP packet to the normal channel.
- the neighbor node device of the device forwards it. This can increase the reliability of message transmission in the event of packet loss in the fast channel.
- the link state change detected by the node 3 is represented by the event A3, and the link detected by the node 8 is used.
- the state change is indicated by event A8.
- the topology change sensing module on the adjacent device node 3 and the node 8 of the node state change detects the topology change event, and the protocol packet generation module generates an LSP message according to the type of the topology change event. After the LSP is generated, the node 3 and the node 8 send the generated LSP packet to the device on the fast channel through the multicast routing table corresponding to the fast channel.
- the convergence processing module of the device continues. The route convergence process is performed on the topology change event. Specifically, the method includes: recalculating the unicast and multicast routing table, and sending the calculated unicast and multicast routing table to the protocol packet sending module.
- the device that receives the LSP packet sent by the fast channel also forwards the LSP packet, and performs convergence processing according to the LSP packet.
- the device on the fast channel of node 3 is node 1
- the devices on the fast channel of node 8 are node 5 and node 2, and node 7 and node 4.
- the node 1 receives only the LSP message corresponding to the event A3, and the node 5 and the node 2, the node 7 and the node 4 only receive the LSP message corresponding to the event A8, only the event
- both A3 and event A8 are received, the nodes in the network can correctly calculate the forwarding routing table to complete the convergence process.
- node 1 node 5 and node 2, node 7 and node 4, also include node 3 and node
- the protocol packet sending module on the device needs to forward the LSP packet received through the fast channel to the neighbor node device of the device through the normal channel. For example, for node 1, pass The fast channel (node 3 -> node 1) receives the LSP packet corresponding to the A3 event. After the convergence process, the protocol packet sending module also needs to receive the LSP through the normal channel (node 1-> node 2).
- node 1 -> node 4 is sent to node 2 and node 4 to ensure that each node in the network can receive both event A3 and event A8, thereby correctly calculating the route forwarding path completion convergence processing. .
- the protocol packet receiving module of the node 2 and the node 4 determines whether the sequence number of the received LSP packet is greater than the sequence number of the packet in the current database.
- the LSP packet is the latest LSP packet.
- the LSP packet sending module sends the LSP packet to the fast channel where the device is located.
- the network routing convergence processing method and apparatus use the pre-established prune-free multicast distribution tree as a fast channel for LSP packet diffusion, and the device in the network uses the text to perform convergence processing.
- the convergence of the entire network is basically not limited by the size of the network and the location of the topology change, and the convergence performance is close to the convergence performance of a single device.
- the storage medium may be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
本发明公开了一种网络路由收敛处理方法,该方法利用预先建立的无剪枝组播分发树作为快速通道,当网络拓扑变化时,拓扑变化源的各邻接设备检测到拓扑变化之后,根据拓扑变化的类型生成 LSP报文,并通过快速通道发送出去,然后这些邻接设备上的收敛处理模块继续针对该拓扑变化事件进行路由收敛处理。快速通道上的设备接收 LSP报文后,将该报文进行转发的同时,也将该报文发送至收敛处理模块进行路由收敛处理。本发明还提供了相应的网络路由收敛处理装置。本发明的方法和装置可以使得整个网络的收敛基本不受网络规模及拓扑变化位置的限制,整个网络的收敛性能接近单台设备的收敛性能。
Description
一种网络路由收敛处理方法和装置 本申请要求于 2011 年 10 月 9 日提交中国专利局、 申请号为 201110302743.0、发明名称为 "一种网络路由收敛处理方法和装置"的中国专 利申请的优先权, 其全部内容通过引用结合在本申请中。 技术领域
本发明涉及通信网络, 尤其涉及一种网络路由收敛处理方法和装置。
背景技术
最短路径桥接 ( Shortest Path Bridging , SPB )技术和多链路透明互连 ( Transparent Interconnection of lots of Links, TRILL )技术均是构建数据中心 大二层网络的技术, 它们都是通过中间系统到中间系统( Intermediate System to Intermediate System, ISIS )协议来实现。 SPB技术和 TRILL技术相对于传 统的二层技术具有优势, 例如均支持等价多路径 (Equal-Cost Multipath , ECMP ), 带宽利用率高, 适合数据中心特定组网等。
对于数据中心来说, 整网的收敛时间是一个很重要的性能指标。 现有的 网络路由收敛处理中, 网络拓朴变化的信息扩散是以波浪的方式逐层扩大。 例如, 某一网络中包括 7个节点设备, 节点设备 1发生故障, 处于同一网络 层的节点 2、3、4、5检测到故障,在检测到故障后通知各自设备的 CPU( Central Processing Unit, 中央处理单元)控制层面进行处理, 在 CPU控制层面处理 完成后, 生成链路状态协议数据单元( Link State Protocol Data Unit, LSP )报 文, 并将生成的 LSP报文通过正常通道发送至外一层的邻居节点 6和 7 , 节 点 6和 7各自的 CPU控制层面根据 LSP报文处理完成后才能够实现网络的收 敛。 这种收敛方式中拓朴变化信息的扩散速度比较慢, 在上述例子中, 网络 收敛的时间至少需要节点 3的 CPU处理时间 tl+协议 4艮文发送时间 t2+节点 6 的 CPU处理时间 t3 , —共为 tl+t2+t3 , 并且网络拓朴越大, 收敛性能越差。 而且, 网络的收敛性能与拓朴变化的位置相关联, 如果拓朴变化发生在网络 的边缘, 则全网的 LSP报文的同步需要横跨整个网络直径, 即使拓朴变化发
生在网络的中心, 全网的 LSP 4艮文的同步也需要横跨网络半径。 发明内容
本发明为解决上述问题而提供了一种网络路由收敛处理方法, 所述方法 包括:
预先建立将网络内各设备连接起来的快速通道, 其中快速通道为无剪枝 的组播分发树;
当该网络的拓朴变化时, 拓朴变化的各邻接设备检测拓朴变化事件, 根 据所述拓朴变化事件的类型生成链路状态协议数据单元 LSP报文, 根据快速 相应地, 本发明还提供了一种网络路由收敛处理装置, 所述装置包括: 预处理模块, 用于预先建立将网络内各设备连接起来的快速通道, 其中 快速通道为无剪枝的组播分发树;
拓朴变化感知模块, 用于检测拓朴变化事件;
协议报文产生模块, 用于根据所述拓朴变化事件的类型生成链路状态协 议数据单元 LSP报文;
协议报文发送模块, 用于根据快速通道对应的组播路由表将 LSP报文进 行发送。
实施本发明实施例, 具有如下有益效果: 通过预先建立的无剪枝的组播 分发树作为快速通道来进行拓朴变化信息的扩散, 由于快速通道上各设备的 收敛处理开始时间相差较小, 基本上处于并行处理的状态, 大大提高了整个 网络的收敛性能, 而且整个网络的收敛时间也基本与网络规模和拓朴变化位 置无关。
附图说明
为了更清楚地说明本发明实施例或现有技术中的技术方案, 下面将对实 施例或现有技术描述中所需要使用的附图作简单地介绍, 显而易见地, 下面 描述中的附图仅仅是本发明的一些实施例, 对于本领域普通技术人员来讲, 在不付出创造性劳动性的前提下, 还可以根据这些附图获得其他的附图。
图。
图 2图示了才艮据本发明实施方式的网络的第一拓朴变化示意图。
图 3图示了才艮据本发明实施方式的网络的拓朴变化示意图。
图 4图示了才艮据本发明实施方式的网络的第二拓朴变化示意图。
图 5 图示了根据本发明实施方式的网络路由收敛处理装置的结构示意 图。
具体实施方式
下面将结合本发明实施例中的附图, 对本发明实施例中的技术方案进行 清楚、 完整地描述, 显然, 所描述的实施例仅仅是本发明一部分实施例, 而 不是全部的实施例。 基于本发明中的实施例, 本领域普通技术人员在没有作 出创造性劳动前提下所获得的所有其他实施例 , 都属于本发明保护的范围。 示意图, 所述方法包括:
S100, 预先建立将网络内各设备连接起来的快速通道, 其中快速通道为 无剪枝的组播分发树;
S102, 当该网络的拓朴变化时, 拓朴变化的各邻接设备检测拓朴变化事 件, 根据所述拓朴变化事件的类型生成链路状态协议数据单元 LSP报文, 根 备。 、、 ; ' ' 、 ' 、、 ' 本发明实施方式中的网络可以运行多链路透明互连 TRILL 协议, 运行 TRILL协议的网络本文称 TRILL网络。本发明实施方式中的网络还可以运行 最短路径桥接 SPB协议, 运行 SPB协议的网络本文称 SPB网络。 本发明提 供的路由收敛处理方法既可以适用于 TRILL网络, 又可以适用于 SPB网络。 需要指出的是, 本发明提供的路由收敛处理方法还可以适用于能够构建无剪 枝快速通道的除上述两种网络的其他网络。 需要说明的是, 本发明实施方式 中的拓朴变化包括链路状态变化和节点设备状态变化, 其中链路状态变化包 括链路故障, 节点设备状态变化包括节点设备故障和节点设置过载。 拓朴变
化为链路故障时, 拓朴变化的各邻接设备是指与发生故障的链路直接连接的 设备, 如果发生故障的链路直接连接的有两台设备, 例如在附图 2所示的网 络架构下, 发生故障的链路直接连接的节点有节点 6和节点 8, 那么本发明 所指的拓朴变化的各邻接设备是指节点 6和节点 8。 如果发生故障的链路直 接连接的有三台设备, 例如在图 3所示的网络架构下, 发生故障的链路直接 连接的节点有节点 1、 节点 2和节点 3 , 那么本发明所指的拓朴变化的各邻接 设备是指节点 1、节点 2和节点 3。在拓朴变化为节点设备故障或设置过载时, 拓朴变化的各邻接设备是指与节点设备直接连接的各设备, 例如, 在附图 4 所示的网络架构下, 节点 6发生故障, 与节点 6直接连接的节点设备为节点 3和节点 8, 那么本发明所指的拓朴变化的邻接设备为节点 3和节点 8, 如果 节点 8发生故障,与节点 8直接相连接的节点设备为节点 5、节点 6和节点 7 , 那么本发明所指的拓朴变化的邻接设备为节点 5、 节点 6和节点 7。
本发明实施方式中, 拓朴变化的各邻接设备根据检测到的拓朴变化生成 LSP报文, 并将生成的 LSP报文通过快速通道转发以通知其他设备该拓朴变 化事件后, 还将继续针对该拓朴变化事件进行路由收敛处理。
S100中的快速通道为无剪枝的组播分发树, 不同的网络可以根据其相应 的协议的具体规定来建立无剪枝的组播分发树作为拓朴变化信息扩散的快速 通道。 在快速通道建立后, 网络中每台设备自身都会包含其所在的快速通道 对应的组播路由表, 在 LSP报文的转发过程中, 可根据设备自身的组播路由 表将报文转发至该组播路由表中对应的下一个路由表项对应的设备, 若该组 播路由表的下一个路由表项为空, 则说明该节点设备是快速通道上的结束节 点, 不需要将 LSP继续转发。 本发明实施方式中可釆用底层芯片来实现组播 路由表的查找, 还可以通过底层芯片根据快速通道对应的组播路由表将 LSP 报文发送至快速通道上的设备, 以节点 1为例, 节点 1的快速通道上的设备 是指节点 1 自身的快速通道对应的组播路由表中对应的设备, 如果节点 1所 在快速通道对应的组播转发表为空, 则说明节点 1为该快速通道上的最后一 个节点, 发送至该节点的 LSP报文无需再继续通过快速通道进行转发。
本发明实施方式中步骤 S102 中根据快速通道对应的组播路由表将生成 的 LSP报文转发至所述快速通道上的设备, 其中所述快速通道上的设备包括
拓朴变化各邻接设备的各自组播路由表中对应的设备。 若拓朴变化的其中一 个邻接设备为快速通道上的最后一个节点设备, 那么该邻接设备查询其自身 的组播路由表即可获知无需将生成的 LSP报文通过快速通道进行转发。 若拓 朴变化的各邻接设备不是快速通道上的最后一个节点设备, 本发明提供的网 络路由收敛处理方法还包括: 所述快速通道上的设备接收所述拓朴变化的各 邻接设备通过快速通道发送的 LSP报文, 根据该快速通道上的设备自身所在 的快速通道对应的组播路由表将所述 LSP 4艮文转发至所述组播路由表对应的 设备, 并将所述 LSP报文发送至该快速通道上的设备的收敛处理模块进行路 由收敛处理, 其中收敛处理模块具体的收敛处理包括, 重新计算单播和组播 路由表, 并将计算得到的单播和组播路由表发送至协议报文发送模块以便进 行报文转发, 本发明的一些实施方式中, 协议报文发送模块可釆用底层芯片 来进行 LSP报文的转发, 底层芯片查找到快速通道对应的组播路由表后, 将 LSP报文向该组播路由表对应设备的出端口进行复制、 发送。 由于底层转发 效率很高, 其转发时间一般远小于路由收敛处理时间, 因而可使得快速通道 上的各设备的路由收敛处理基本上处于并行处理的状态。
对于 TRILL网络而言, 快速通道建立的步骤可包括:
完成 TRILL网络部署之后, 根据 ISIS协议完成邻居建立和数据库同步; 运行 TRILL协议的设备中树根优先级最高的第一运行 TRILL协议的设备 优先指定树根;
第一运行 TRILL协议的设备指定建立组播树的最大数目 K和 S个组播树 树根, 其中,
当 S等于 K时, 所述 S个组播树树根作为所述快速通道的树根, 当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先级进行排 序, 前 K个设备作为所述快速通道的树根,
当 S小于 K时,将所述 S个组播树树根作为所述快速通道的前 S个树根, 将网络中除所述 S个组播树树根之外的设备按树根优先级排序后的前 K-S个 设备作为所述快速通道的 K-S个树根。
具体实施方式中, TRILL网络部署完成之后, 运行 TRILL协议的设备根 据 ISIS协议完成邻居建立和数据库同步,最后网络内每台运行 TRILL协议的
设备都拥有整个网络的拓朴信息。 每台运行 TRILL协议的设备可以通过 LSP 报文发布它能够计算的组播树的最大数目。 网络中树根优先级(Tree Root Priority ) 最高的第一运行 TRILL 协议的设备可以优先指定树根。 第一运行 TRILL协议的设备指定网络内可以建立的组播树的最大数目 K和 S个组播树 树才艮 (例^口 nickname 1、 nickname2、 nickname3... nickname S )。 当 S等于 K时, 第一运行 TRILL协议的设备指定的 S个组播树树根作为快速通道的树根; 当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先级进行排序, 前 K个作为快速通道的树根; 当 S小于 K时,将第一运行 TRILL协议的设备指 定的 S个组播树树根作为快速通道的前 S个树根, 将网络中除指定的树根之 外的设备按树根优先级排序, 将排序后的前 K-S个设备作为快速通道的 K-S 个树根。 例如第一运行 TRILL协议的设备指定 K=4, S=2{Tx, Ty} , 其中, Tx和 Ty为第一运行 TRILL协议的设备指定的组播树树根, 由于第一运行 TRILL协议的设备指定的组播树的最大数目 K为 4, 那么还需要将网络内除 了 Tx和 Ty之外的设备按照树根优先级排序后取前(K-S=4-2=2 )个设备作 为快速通道的另外 2个树根, 若网络还包括设备 Ta、 Tb和 Tc, 网络中树根 优先级的排序顺序为 Ty>Ta>Tc>Tb>Tx,则 K颗树的树根为 { Tx, Ty, Ta, Tc}。 每台设备都基于通过上述计算得到的树根独立计算无剪枝组播分发树作为快 通过每台设备上存储的组播路由表将所有的设备在转发层面上连通, 相当于 一个虚拟的以太接口将网络内各个设备互相连接。
对于 SPB网络而言,网络内的任意节点设备都可以作为快速通道的树根。 具体地, 可以根据相应的协议规定来建立无剪枝的组播分发树作为拓朴变化 信息扩散的快速通道。
在通过上述的组播分发树作为快速通道转发流量之前, SPB 网络或 TRILL 网络内的设备需要对设备进行邻居检查和 RPF ( Reverse Path Forwarding, 反向路径转发)检查以防止出现环路。 LSP报文通过快速通道发 送时, LSP报文作为原始数据报文封装在 TRILL协议报文或 SPB协议报文的 内层, 作为 TRILL数据报文或 SPB数据报文封装的数据载荷。
本发明的实施方式中, 网络的拓朴变化可以包括链路状态变化,例如链路
up/down, 开销 (cost )值变化等, 还可以包括节点设备状态变化, 例如节点 设备故障、 设置过载等, 设置过载是指网络管理员需要针对某个节点进行升 级或维护时, 将该节点设置成过载状态, 该节点会通过 LSP报文通知全网除 该节点外的其他节点设备本节点出现过载信息, 让其他节点在计算路由表时 将该节点绕开。
参见图 2 , 图示了根据本发明实施方式的网络的第一拓朴变化示意图, 网络中包括 8个节点设备, 节点 8为树根。 如图 2所示, 节点设备 6和节点 设备 8之间的链路发生变化。 预先建立的快速通道的树根为节点 8, 以节点 8 为树根的各个支路分别为 {节点 5 , 节点 2}, {节点 6, 节点 3 , 节点 1 } , {节点 7 , 节点 4}。 在预先建立的上述快速通道下, 釆用本发明的网络路由收敛的 处理方法进行处理的具体步骤可以包括:
链路状态变化的邻接设备节点 6和节点 8检测拓朴变化事件;
节点 6和节点 8根据所述拓朴变化事件类型生成 LSP报文;
节点 6和节点 8通过快速通道对应的组播路由表将生成的 LSP报文发送 至所述快速通道上的设备, 然后节点 6和节点 8继续针对该拓朴变化事件进 行路由收敛处理, 具体地, 包括重新计算单播和组播路由表, 并将计算得到 的单播和组播路由表并发送至协议报文发送模块;
节点 6和节点 8快速通道上的设备接收通过快速通道发送的 LSP报文, 通过快速通道对应的组播路由表将所述 LSP 4艮文转发至所述组播路由表对应 的设备, 还将所述接收到的 LSP报文发送至设备的收敛处理模块以便进行收 敛处理, 具体地, 包括重新计算单播和组播路由表, 将计算得到的路由表发 送至协议报文发送模块以便进行报文转发, 其中,
节点 6和节点 8以及节点 6和节点 8的快速通道上的设备在设备自身完 成收敛处理之后, 还将所述 LSP报文通过正常通道发送至设备自身的邻居节 点。
本发明的具体实施方式中, 链路状态发生变化后, 链路状态变化的邻接 设备节点 6和节点 8检测拓朴变化事件, 即: 链路状态变化, 根据拓朴变化 事件的类型 (例如链路 up/down、 cost值变化等)生成 LSP报文, 一方面, 节点 6和节点 8将生成的 LSP报文通过快速通道对应的组播路由表发送至快
速通道上的设备, 对于节点 6而言, 其所在的快速通道对应的组播路由表的 下一表项对应的节点设备为节点 3 , 而节点 3所在的快速通道对应的组播路 由表的下一表项对应的节点设备为节点 1 , 在附图 2所示的实施方式中, 节 点 1所在的快速通道对应的组播路由表的下一表项为空, 节点 1在接收到通 过快速通道发送的 LSP报文后, 查找该节点 1 自身所在的快速通道对应的组 播路由表后获知其组播路由表的下一表项为空, 无需再将 LSP报文通过快速 通道继续转发。 对于节点 8而言, 其所在的快速通道上的设备为节点 7和节 点 4, 以及节点 5和节点 2, 其通过快速通道转发 LSP "^文的过程类似, 在 此不再赘述。 各节点设备转发的具体步骤包括通过底层芯片查找设备自身所 在的快速通道对应的组播路由表, 根据查找到的组播路由表向该组播路由表 中对应设备的出端口进行复制、 发送; 另一方面, 节点 6和节点 8继续针对 该拓朴变化事件进行路由收敛处理, 收敛处理的具体步骤包括新计算单播和 组播路由表, 并将计算得到的单播和组播路由表发送至协议报文发送模块以 便进行报文转发。 快速通道中扩散的 LSP报文是根据控制协议本身生成的 LSP报文, 因而可以釆用协议本身的消息摘要算法进行安全处理, 从而避免 网络攻击, 保证网络的安全性。 在如图 2所示的实施方式中, 节点 6和节点 8的快速通道上的设备分别对应的节点 3和节点 1、 节点 7和节点 4、 节点 5 和节点 2, 由于各个节点设备的处理是类似的, 下面以节点 3为例说明设备 对接收到的快速通道上转发的 LSP报文的处理过程, 节点 3在接收到通过快 速通道发送的 LSP报文后, 一方面, 通过底层芯片查找快速通道对应的组播 路由表, 并根据查找到的组播路由表将接收到的 LSP报文向该组播路由表对 应的设备的出端口进行复制和发送; 另一方面, 节点 3在接收到通过快速通 道发送的 LSP报文后,还会将所接收到的 LSP报文发送至节点 3的收敛处理 模块进行收敛处理, 具体地, 包括计算单播和组播路由表, 并将计算得到的 单播和组播路由表发送至协议报文发送模块以便进行报文转发, 其中重新计 算的组播路由表可以用于更新原有的快速通道对应的组播路由表。 在网络内 的所有设备都接收到拓朴变化的 LSP报文,并根据接收到的 LSP完成收敛处 理后, 整个网络完成收敛。 值得指出的是, 本发明实施方式中网络内的设备 通过快速通道对 LSP报文进行转发, 一般的传递时间在微秒级别, 通常远小
于收敛处理的时间 , 相对于现有的网络的收敛时间几百毫秒 , 该时间基本可 以忽略不计, 换而言之, 各设备收敛处理的开始时间仅相差在微秒级别, 基 本上处于并行处理的状态, 这使得整个网络的收敛时间基本上等于单个节点 设备的收敛时间, 而且整个网络的收敛基本不再受网络规模和拓朴变化位置 的限制, 大大地提高了网络的收敛性能。
进一步地, 本发明的实施方式中, 节点 6和节点 8以及节点 6和节点 8 快速通道上的设备在设备自身收敛处理完成后, 可以通过正常通道将 LSP报 文向与其连接的邻居节点进行转发。 这样在快速通道中报文丟失的情况下, 可以增加报文传送的可靠性。 本文所述的正常通道是指网络中设备之间已经 存在的报文传输通道, 在该通道中 LSP报文进行串行传输, 在此称之为 "正 通通道,,。
需要指出的是, 网络中的设备可能从正常通道和快速通道收到重复的 LSP报文, 网络各自的协议可以自动处理这种重复发送的情况, 在重复发送 的情况下, 对重复接收到的报文不作出处理。
当网络内所有设备都接收到拓朴变化的 LSP报文并进行相应的收敛处理 后, 整个网络完成收敛。 网络收敛后重新计算新的组播分发树, 为后续的新 的拓朴变化提供新的快速通道。
参见图 4 , 图示了根据本发明实施方式的网络的第二拓朴变化示意图。 网络中包括 8个节点设备, 节点 8为树根。 如图 3所示, 节点设备 6状态发 生变化, 可能是节点故障或者设置过载等。 预先建立的快速通道的树根为节 点 8, 以节点 8为树根的各个支路分别为 {节点 5 , 节点 2}, {节点 6 , 节点 3 , 节点 1 } , {节点 7 , 节点 4}。 在图 3所示的网络架构下, 节点 6状态发生变化 涉及的链路变化包括节点 3和节点 6之间和节点 8和节点 6之间两条链路的 状态变化。 在预先建立的上述快速通道下, 釆用本发明的网络路由收敛的处 理方法进行处理的具体步骤可以包括:
节点状态变化的邻接设备节点 3和节点 8检测拓朴变化事件;
节点 3和节点 8根据所述拓朴变化事件的类型生成 LSP报文;
节点 3和节点 8通过快速通道对应的组播路由表将生成的 LSP报文发送
至所述快速通道上的设备, 然后节点 3和节点 8继续针对所述拓朴变化事件 进行路由收敛处理, 具体地, 包括重新计算单播和组播路由表, 并将计算得 到的单播和组播路由表发送至协议报文发送模块;
所述快速通道上的设备接收通过快速通道发送的 LSP "^文, 通过快速通 道对应的组播路由表将所述 LSP报文转发至所述组播路由表对应的设备, 并 将所述接收到的 LSP报文发送至设备的收敛处理模块进行收敛处理 ,具体地 , 包括重新计算单播和组播路由表, 将计算得到的路由表发送至协议报文发送 模块以便进行报文转发;
所述快速通道上的设备在设备自身收敛处理完成后, 还将所述 LSP报文 通过正常通道发送至设备自身的邻居节点;
所述快速通道上的设备的邻居节点设备接收通过正常通道转发的 LSP报 备所在的快速通道上的各设备进行转发。
本发明具体实施方式中, 节点 6状态变化后, 节点 3和节点 8检测到与 节点 6相关的链路的状态变化事件, 为描述方便, 将节点 3检测到的链路状 态变化用事件 A3表示, 将节点 8检测到的链路状态变化用事件 A8表示。 节 点 3和节点 8分别检测链路状态变化事件 A3和 A8 , 根据链路状态变化的类 型生成相应的 LSP 4艮文, 一方面, 节点 3和节点 8将生成的 LSP ^ l通过快 速通道对应的组播路由表发送至快速通道上的设备, 具体地, 通过底层芯片 查找设备自身所在的快速通道对应的组播路由表, 并根据查找到的组播路由 表将接收到的 LSP报文向该组播路由表对应设备的出端口进行复制和发送; 另一方面, 节点 3和节点 8继续针对该拓朴变化事件进行路由收敛处理, 具 体地, 包括重新计算单播和组播路由表, 并将计算得到的单播和组播路由表 并发送至协议报文发送模块, 其中重新计算的组播路由表可以用于更新原有 的快速通道对应的组播路由表。 在如图 4所示的网络架构下, 节点 3快速通 道上的设备为节点 1 , 节点 8快速通道上的设备为节点 5和节点 2, 以及节点 7和节点 4。 在上述 LSP报文通过快速通道的扩散中, 节点 1仅收到事件 A3 对应的 LSP报文, 节点 5和节点 2, 节点 7和节点 4仅收到事件 A8对应的 LSP报文, 只有事件 A3和事件 A8都收到时, 网络中的节点才可以正确地计
算转发路由表完成收敛处理。 在上述设备(节点 1、 节点 5和节点 2、 节点 7 和节点 4, 也包括节点 3和节点 8 ) 自身完成收敛处理后, 还需要通过正常通 道将通过快速通道接收到的 LSP报文向设备自身的邻居节点设备进行转发。 例如, 对于节点 1 , 在通过快速通道 (节点 3->节点 1 )接收到 A3事件对应 的 LSP报文,进行收敛处理后,还需要将接收到的 LSP报文通过正常通道(节 点 1->节点 2, 节点 1->节点 4 )发送至节点 2和节点 4, 以保证网络内的各个 节点既能够收到事件 A3又能够收到事件 A8, 从而能够正确地计算路由转发 路径完成收敛处理。 节点 2和节点 4在接收到通过正常通道发送的 LSP报文 后, 判断接收到的 LSP报文的报文序列号是否大于其自身当前数据库的报文 序列号, 如果大于, 则说明此 LSP报文是最新的 LSP报文, 需要通过其所在 的快速通道进行发送。通过上述的再次转发可以保证事件 A3和事件 A8都可 以被网络内的节点收到, 从而正确地计算出新的路由转发路径, 为后续的新 的拓朴变化提供新的快速通道。 体地说明, 下面将结合附图和实施例对本发明提供的网路路由收敛处理装置 进行具体地说明。
参见图 5 , 图示了根据本发明实施方式的网络路由收敛处理装置的结构 示意图。 网路路由收敛处理装置 500可包括:
预处理模块 502 , 用于预先建立将网络内各设备连接起来的快速通道, 其中快速通道为无剪枝的组播分发树;
拓朴变化感知模块 504, 用于检测拓朴变化事件;
协议报文产生模块 506 , 用于根据所述拓朴变化事件的类型生成链路状 态协议数据单元 LSP报文;
协议报文发送模块 508 , 用于根据快速通道对应的组播路由表将 LSP报 文进行发送。
本发明的实施方式中, 网络路由收敛处理装置 500除了包括上述模块外, 还可以包括:
收敛处理模块 510, 用于进行路由收敛处理;
协议报文接收模块 512, 用于接收 LSP报文;
查找处理模块 514, 用于为接收到的 LSP报文查找组播路由表, 还用于 将接收到的 LSP报文发送至收敛处理模块以便进行收敛处理。
本发明实施方式提供的网络路由收敛处理装置的拓朴变化感知模块、 协 议报文产生模块、 协议报文发送模块、 协议报文接收模块、 收敛处理模块以 及查找处理模块可配置在网络内的各个设备上, 本发明实施方式中网络内的 各个设备处于对等的地位, 可以在各个设备上配置同等的功能模块。 在网络 拓朴变化的收敛处理过程中, 可以只使用部分设备的部分功能模块。
本发明的一些实施方式中, 查找处理模块和协议报文发送模块可釆用底 层芯片来实现, 其中底层芯片可以釆用本领域技术人员已知的能够实现本发 明所述功能的任何芯片。
本发明实施方式中的网络可以运行多链路透明互连 TRILL协议, 本发明 中将运行 TRILL协议的网络称为 TRILL网络。本发明实施方式中的网络还可 以运行最短路径桥接 SPB协议, 本发明中将运行 SPB协议的网络称为 SPB 网络。 本发明提供的路由收敛处理装置既可以适用于 TRILL网络, 又可以适 用于 SPB网络。 需要指出的是, 本发明提供的路由收敛处理装置还可以适用 于能够构建无剪枝快速通道的除上述两种网络的其他网络。
本发明实施方式中的快速通道为无剪枝组播分发树, 不同的网络可以根 据其相应的协议的具体规定来建立无剪枝的组播分发树作为拓朴变化信息扩 散的快速通道。 快速通道相当于一个虚拟的以太接口将网络内各个设备互相 连接。 在通过建立的组播分发树作为快速通道转发流量之前, SPB 网络或 TRILL网络内的设备需要对设备进行邻居检查和 RPF检查以防止环路出现。
LSP报文通过快速通道发送时, LSP报文作为原始数据报文封装在 TRILL协 议报文或 SPB协议报文的内层。
对于 TRILL网络, 本发明实施方式中的网络路由收敛处理装置中的预处 理模块 402具体可以用于在完成 TRILL网络部署之后, 根据中间系统到中间 系统 ISIS协议完成邻居建立和数据库同步,运行 TRILL协议的设备中树根优 先级最高的第一运行 TRILL协议的设备优先指定树根,第一运行 TRILL协议 的设备指定建立组播树的最大数目 K和 S个组播树树根, 其中,
当 S等于 K时, 所述 S个组播树树根作为所述快速通道的树根,
当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先级进行排 序, 前 K个设备作为所述快速通道的树根,
当 S小于 K时,将所述 S个组播树树根作为所述快速通道的前 S个树根, 将网络中除所述 S个组播树树根之外的设备按树根优先级排序后的前 K-S个 设备作为所述快速通道的 K-S个树根。
本发明具体实施方式中, 预处理模块在 TRILL网络部署完成之后, 根据 中间系统到中间系统 ISIS协议完成邻居建立和数据库同步, 最后网络内每台 运行 TRILL协议的设备都拥有整个网络的拓朴信息。每台运行 TRILL协议的 设备可以通过 LSP报文发布它能够计算的组播树的最大数目。运行 TRILL协 议的设备中树根优先级最高的第一运行 TRILL 协议的设备可以优先指定树 根。 第一运行 TRILL协议的设备指定网络内可以建立的组播树的最大数目 K 和 S个组播树树根 (例如 nickname 1、 nickname2、 nickname3... nickname S )。 当 S等于 K时,第一运行 TRILL协议的设备指定的 S个组播树树根作为快速 通道的树根; 当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先 级进行排序,前 K个作为快速通道的树根;当 S小于 K时,将第一运行 TRILL 协议的设备指定的 S个组播树树根作为快速通道的前 S个树根, 将网络中除 指定的树根之外的设备按树根优先级排序, 将排序后的前 K-S个设备作为快 速通道的 K-S个树根。 例如第一运行 TRILL协议的设备指定 K=4, S=2{Tx, Ty} , 其中, Tx和 Ty为第一运行 TRILL协议的设备指定的组播树树根, 由 于第一运行 TRILL协议的设备指定的组播树的最大数目 K为 4, 那么还需要 将网络内除了 Tx和 Ty之外的设备按照树根优先级排序后取前(K-S=4-2=2 ) 个设备作为快速通道的另外 2个树根, 若网络还包括设备 Ta、 Tb和 Tc, 网 络中树根优先级的排序顺序为 Ty>Ta>Tc>Tb>Tx, 则 K颗树的树根为 { Tx, Ty, Ta, Tc}。每台设备都基于通过上述计算得到的树根独立计算组播分发树作 体地, 通过每台设备上存储的组播路由表将所有的设备在转发层面上连通。
对于 SPB网络而言,网络内的任意节点设备都可以作为快速通道的树根。 相应的预处理模块可根据协议的具体规定建立无剪枝的组播分发树作为拓朴 变化信息扩散的快速通道。
本发明的实施方式中, 网络的拓朴变化可以包括链路状态变化,例如链路 up/down, cost值变化等,还可以包括节点设备状态变化,例如节点设备故障、 设置过载等,设置过载是指网络管理员需要针对某个节点进行升级或维护时, 将该节点设置成过载状态, 该节点会通过 LSP报文通知全网除该节点外的其 他节点设备本节点出现过载信息,让其他节点在计算路由表时将该节点绕开。
在如图 2所示的拓朴变化下, 链路状态变化的邻接设备节点 6和节点 8 上的拓朴变化感知模块检测拓朴变化事件, 协议报文产生模块根据所述拓朴 变化的类型生成链路状态协议数据单元 LSP报文, 在生成 LSP报文之后, 一 方面, 查找设备自身所在的快速通道对应的组播路由表, 并根据查找到的组 播路由表将 LSP报文进行转发; 另一方面, 设备自身的收敛处理模块继续针 对该拓朴变化事件进行路由收敛处理, 具体地, 包括重新计算单播和组播路 由表, 将计算得到的单播和组播路由表发送至协议报文发送模块以便进行报 文转发, 其中计算得到的组播路由表可更新原有的快速通道对应的组播路由 表。 节点 3的协议报文接收模块接收节点 6的协议报文发送模块通过快速通 道转发的 LSP报文, 节点 3在接收到 LSP报文后, 一方面查找设备自身所在 的快速通道的组播路由表, 根据查找到的组播路由表将 LSP报文进行转发, 另一方面通过收敛处理模块根据所生成 LSP报文进行路由收敛处理。 对于节 点 5和节点 7处理过程是类似的, 在此不再赘述。 节点 3将接收到的 LSP报 文继续转发至节点 1 , 同理节点 5和节点 7将接收到的 LSP报文继续分别转 发至节点 2和节点 4。需要说明的是,本发明实施方式中釆用快速通道将 LSP 报文进行转发, 处理时间一般在微秒级别, 换而言之, 各个设备接收到 LSP 报文的时间相差是微秒级别, 远小于设备的收敛处理时间, 各个设备对 LSP 报文的处理基本上可看作是并行的, 这使得整个网络的收敛性能接近单台设 备的收敛性能, 而且整个网络的收敛基本不受网络规模和拓朴变化位置的限 制。
进一步地, 链路状态变化的邻接设备以及接收到通过快速通道发送的 LSP 的设备自身完成收敛后, 所述设备的协议报文发送模块还可用于通过正 常通道将接收到的 LSP报文向该设备自身的邻居节点设备进行转发。 这可以 在快速通道中出现报文丟失的情况下, 增加报文传送的可靠性。
本发明的实施方式中, 当在如图 4所示的拓朴状态变化下, 在此为描述 方便, 将节点 3检测到的链路状态变化用事件 A3表示, 将节点 8检测到的 链路状态变化用事件 A8表示。 节点状态变化的邻接设备节点 3和节点 8上 的拓朴变化感知模块检测拓朴变化事件, 协议报文产生模块根据拓朴变化事 件的类型生成 LSP报文。 在生成 LSP报文后, 节点 3和节点 8, —方面, 将 生成的 LSP报文通过快速通道对应的组播路由表发送至快速通道上的设备, 另一方面, 设备自身的收敛处理模块继续针对所述拓朴变化事件进行路由收 敛处理, 具体地, 包括重新计算单播和组播路由表, 将计算得到的单播和组 播路由表发送至协议报文发送模块。 接收到通过快速通道发送的 LSP报文的 设备也是一方面将 LSP报文进行转发, 一方面根据 LSP报文进行收敛处理。 在如图 4所示的网络架构下, 节点 3快速通道上的设备为节点 1 , 节点 8快 速通道上的设备为节点 5和节点 2, 以及节点 7和节点 4。 在上述 LSP "^文 通过快速通道的扩散中, 节点 1仅收到事件 A3对应的 LSP报文, 节点 5和 节点 2, 节点 7和节点 4仅收到事件 A8对应的 LSP报文, 只有事件 A3和事 件 A8都收到时, 网络中的节点才可以正确地计算转发路由表完成收敛处理。 在上述设备(节点 1、 节点 5和节点 2、 节点 7和节点 4, 也包括节点 3和节 点 8 ) 自身完成收敛处理后, 还需要利用设备上的协议报文发送模块通过正 常通道将通过快速通道接收到的 LSP报文向设备自身的邻居节点设备进行转 发。 例如, 对于节点 1 , 在通过快速通道(节点 3->节点 1 )接收到 A3事件 对应的 LSP报文,进行收敛处理后,还需要协议报文发送模块将接收到的 LSP "^文通过正常通道(节点 1->节点 2,节点 1->节点 4 )发送至节点 2和节点 4, 以保证网络内的各个节点既能够收到事件 A3又能够收到事件 A8, 从而能够 正确地计算路由转发路径完成收敛处理。 节点 2和节点 4的协议报文接收模 块在接收到通过正常通道发送的 LSP报文后,判断接收到的 LSP报文的报文 序列号是否大于其自身当前数据库的报文序列号, 如果大于, 则说明此 LSP 报文是最新的 LSP报文,需要通过协议报文发送模块将此 LSP报文向该设备 自身所在的快速通道进行发送。 通过上述的再次转发可以保证事件 A3 和事 件 A8 都可以被网络内的节点收到, 从而正确地计算出新的路由转发路径, 为后续的新的拓朴变化提供新的快速通道。
综上所述, 本发明提供的网络路由收敛处理方法和装置, 将预先建立的 无剪枝的组播分发树作为 LSP报文扩散的快速通道, 而且网络内的设备釆用 文进行收敛处理, 使得整个网络的收敛基本不受网络规模和拓朴变化位置的 限制, 收敛性能接近单台设备的收敛性能。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流 程, 是可以通过计算机程序来指令相关的硬件来完成, 所述的程序可存储于 一计算机可读取存储介质中, 该程序在执行时, 可包括如上述各方法的实施 例的流程。其中,所述的存储介质可为磁碟、光盘、只读存储记忆体( Read-Only Memory, ROM )或随机存 己忆体 ( Random Access Memory, RAM )等。
以上所揭露的仅为本发明的较佳实施例而已, 当然不能以此来限定本发 明之权利范围, 因此依本发明权利要求所作的等同变化, 仍属本发明所涵盖 的范围。
Claims
1、 一种网络路由收敛处理方法, 其特征在于, 所述方法包括: 预先建立将网络内各设备连接起来的快速通道, 其中快速通道为无剪 枝的组播分发树;
当该网络的拓朴变化时, 拓朴变化的各邻接设备检测拓朴变化事件, 根据所述拓朴变化事件的类型生成链路状态协议数据单元 LSP报文, 根据 快速通道对应的组播路由表将生成的 LSP报文发送至所述快速通道上的设 备。
2、 如权利要求 1所述的方法, 其特征在于, 所述方法还包括: 所述快速通道上的设备接收所述 LSP报文, 根据该设备自身所在的快 速通道对应的组播路由表将所述 LSP 文转发至所述组播路由表对应的设 备, 并根据所述 LSP报文进行路由收敛处理。
3、 如权利要求 1或 2所述的方法, 其特征在于, 所述网络运行多链路 透明互连 TRILL协议。
4、 如权利要求 1或 2所述的方法, 其特征在于, 所述网络运行最短路 径桥接 SPB协议。
5、 如权利要求 3所述的方法, 其特征在于, 所述快速通道建立的步骤 包括:
完成 TRILL网络部署之后,运行 TRILL协议的设备根据中间系统到中 间系统 ISIS协议完成邻居建立和数据库同步;
运行 TRILL协议的设备中树根优先级最高的第一运行 TRILL协议的设 备优先指定树根;
第一运行 TRILL协议的设备指定建立组播树的最大数目 K和 S个组播 树树根, 其中, 当 S等于 K时, 所述 S个组播树树根作为所述快速通道的树根, 当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先级进行 排序, 前 K个设备作为所述快速通道的树根,
当 S小于 K时, 将所述 S个组播树树根作为所述快速通道的前 S个树 根, 将网络中除所述 S个组播树树根之外的设备按树根优先级排序后的前 κ-s个设备作为所述快速通道的 κ-s个树根。
6、 如权利要求 5所述的方法, 其特征在于, 所述网络的拓朴变化为链 路状态变化时, 所述方法包括:
所述链路状态变化的各邻接设备检测拓朴变化事件, 根据所述拓朴变 化事件的类型生成 LSP 报文, 根据快速通道对应的组播路由表将生成的 LSP 报文发送至所述快速通道上的设备, 然后所述链路状态变化的各邻接 设备继续针对该拓朴变化事件进行路由收敛处理,
通过发送所述 LSP报文, 使得所述快速通道上的设备接收到所述 LSP 报文, 根据该快速通道上的设备自身所在的快速通道对应的组播路由表将 所述 LSP报文转发至所述组播路由表对应的设备, 并根据所述 LSP报文进 行收敛处理, 其中,
所述链路状态变化的各邻接设备和所述快速通道上的设备在设备自身 完成收敛之后, 还将所述 LSP报文通过正常通道发送至设备自身的邻居节 点。
7、 如权利要求 5所述的方法, 其特征在于, 所述网络的拓朴变化为节 点设备状态变化时, 所述方法包括:
所述节点设备状态变化的各邻接设备检测拓朴变化事件, 根据所述拓 朴变化事件的类型生成 LSP报文, 根据快速通道对应的组播路由表将生成 的 LSP报文发送至所述快速通道上的设备, 然后所述节点设备状态变化的 各邻接设备继续针对该拓朴变化事件进行路由收敛处理,
通过发送所述 LSP报文, 使得所述快速通道上的设备接收到所述 LSP 报文, 根据该快速通道上的设备自身所在的快速通道对应的组播路由表将 所述 LSP报文转发至所述组播路由表对应的设备, 并根据所述 LSP报文进 行收敛处理, 然后该设备在设备自身完成收敛处理后, 还将所述 LSP报文 通过正常通道发送至设备自身的邻居节点以便所述邻居节点根据所述 LSP 报文的报文序列号和所述邻居节点自身所在的快速通道对应的组播路由表 将所述 LSP进行转发。
8、 如权利要求 7所述的方法, 其特征在于, 所述节点设备状态变化包 括节点故障和设置过载。
9、 一种网络路由收敛处理装置, 其特征在于, 所述装置包括: 预处理模块, 用于预先建立将网络内各设备连接起来的快速通道, 其 中快速通道为无剪枝的组播分发树;
拓朴变化感知模块, 用于检测拓朴变化事件;
协议报文产生模块, 用于根据所述拓朴变化事件的类型生成链路状态 协议数据单元 LSP报文;
协议报文发送模块, 用于根据快速通道对应的组播路由表将 LSP报文 进行发送。
10、 如权利要求 9所述的装置, 其特征在于, 所述装置还包括: 收敛处理模块, 用于进行路由收敛处理;
协议报文接收模块, 用于接收 LSP报文;
查找处理模块, 用于为接收到的 LSP报文查找组播路由表, 还用于将 接收到的 LSP报文发送至收敛处理模块以便进行收敛处理。
11、 如权利要求 9或 10所述的装置, 其特征在于, 所述预处理模块用 于在完成 TRILL网络部署之后,根据中间系统到中间系统 ISIS协议完成邻 居建立和数据库同步, 运行 TRILL协议的设备中树根优先级最高的第一运 行 TRILL协议的设备优先指定树根,第一运行 TRILL协议的设备指定建立 组播树的最大数目 K和 S个组播树树根, 其中, 当 S等于 K时, 所述 S个组播树树根作为所述快速通道的树根, 当 S等于零时, 将网络内运行 TRILL协议的设备按照树根优先级进行 排序, 前 K个设备作为所述快速通道的树根,
当 S小于 K时, 将所述 S个组播树树根作为所述快速通道的前 S个树 根, 将网络中除所述 S个组播树树根之外的设备按树根优先级排序后的前 κ-s个设备作为所述快速通道的 κ-s个树根。
12、 如权利要求 11所述的装置, 其特征在于, 所述网络的拓朴变化为 链路状态变化时,
所述协议报文发送模块, 还用于在设备自身完成收敛处理后, 将所述 LSP报文通过正常通道发送至设备自身的邻居节点。
13、 如权利要求 11所述的装置, 其特征在于, 所述网络的拓朴变化为 节点设备状态变化时,
所述协议报文发送模块, 还用于在设备自身完成收敛处理后, 将所述 LSP报文通过正常通道发送至设备自身的邻居节点;
所述协议报文接收模块,还用于根据接收到的 LSP报文的报文序列号来 确定是否通过协议报文发送模块向设备自身所在的快速通道上的设备进行 转发。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201110302743.0 | 2011-10-09 | ||
CN201110302743.0A CN103036787B (zh) | 2011-10-09 | 2011-10-09 | 一种网络路由收敛处理方法和装置 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2013053276A1 true WO2013053276A1 (zh) | 2013-04-18 |
Family
ID=48023291
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2012/081571 WO2013053276A1 (zh) | 2011-10-09 | 2012-09-19 | 一种网络路由收敛处理方法和装置 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN103036787B (zh) |
WO (1) | WO2013053276A1 (zh) |
Families Citing this family (13)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104113483B (zh) * | 2013-04-19 | 2017-06-20 | 华为技术有限公司 | 一种组播通信方法和汇聚交换机 |
CN104753753B (zh) * | 2013-12-31 | 2018-11-16 | 杭州华为数字技术有限公司 | 一种qpi报文的传输方法、设备及计算机系统 |
CN105282041A (zh) * | 2014-07-18 | 2016-01-27 | 中兴通讯股份有限公司 | 基于isis的洪泛方法及装置 |
CN104270315B (zh) * | 2014-09-03 | 2017-08-01 | 武汉烽火网络有限责任公司 | Trill广播网络中组播分发树的计算方法及系统 |
CN104506430A (zh) * | 2015-01-15 | 2015-04-08 | 安徽皖通邮电股份有限公司 | 一种isis协议lsp洪泛和同步方法 |
CN105634952B (zh) * | 2015-07-29 | 2019-01-22 | 新华三技术有限公司 | Lsp报文快速通告方法以及装置 |
CN105656781B (zh) * | 2016-03-17 | 2019-06-21 | 华为技术有限公司 | 一种链路状态数据包的传输方法及网络设备 |
CN108111423B (zh) * | 2017-12-28 | 2020-11-17 | 迈普通信技术股份有限公司 | 流量传输管理方法、装置及网络分路设备 |
CN110365585B (zh) * | 2018-03-26 | 2021-08-03 | 武汉大学 | 一种基于多代价指标的路由裁剪优化方法 |
CN109039702B (zh) * | 2018-06-26 | 2022-03-29 | 成都鼎桥通信技术有限公司 | 专网集群系统中组播组网的实现方法和装置 |
CN108965141B (zh) * | 2018-09-18 | 2021-03-30 | 深圳市风云实业有限公司 | 一种多路径路由树的计算方法及装置 |
CN113810274A (zh) * | 2020-06-16 | 2021-12-17 | 华为技术有限公司 | 一种路由处理方法及相关设备 |
CN112787939B (zh) * | 2020-12-30 | 2022-11-08 | 迈普通信技术股份有限公司 | 路径转发表更新方法及装置 |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070127395A1 (en) * | 2005-12-07 | 2007-06-07 | Cisco Technology, Inc. | Preventing transient loops in broadcast/multicast trees during distribution of link state information |
US20070253416A1 (en) * | 2006-04-28 | 2007-11-01 | Raj Alex E | Method and apparatus for forwarding label distribution protocol multicast traffic during fast reroute |
CN101330448A (zh) * | 2007-06-21 | 2008-12-24 | 华为技术有限公司 | 一种通告链路状态信息及确定组播转发路径的方法及装置 |
CN101453385A (zh) * | 2007-11-30 | 2009-06-10 | 华为技术有限公司 | 一种故障通告的方法及设备 |
CN101521927A (zh) * | 2009-04-03 | 2009-09-02 | 中兴通讯股份有限公司 | 一种组播转发路径收敛的方法和系统 |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN101192959B (zh) * | 2006-11-28 | 2011-04-20 | 中兴通讯股份有限公司 | 一种自动交换光网络组播业务连接的恢复方法 |
CN101192957A (zh) * | 2006-11-28 | 2008-06-04 | 中兴通讯股份有限公司 | 一种自动交换光网络组播业务路由的计算方法 |
CN101192956A (zh) * | 2006-11-28 | 2008-06-04 | 中兴通讯股份有限公司 | 一种自动交换光网络组播业务组播树的计算方法 |
CN101335689B (zh) * | 2007-06-26 | 2011-11-02 | 华为技术有限公司 | 跟踪路由的实现方法及设备 |
CN101163103A (zh) * | 2007-11-07 | 2008-04-16 | 孙先花 | 一种实现快速重路由的方法 |
-
2011
- 2011-10-09 CN CN201110302743.0A patent/CN103036787B/zh active Active
-
2012
- 2012-09-19 WO PCT/CN2012/081571 patent/WO2013053276A1/zh active Application Filing
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20070127395A1 (en) * | 2005-12-07 | 2007-06-07 | Cisco Technology, Inc. | Preventing transient loops in broadcast/multicast trees during distribution of link state information |
US20070253416A1 (en) * | 2006-04-28 | 2007-11-01 | Raj Alex E | Method and apparatus for forwarding label distribution protocol multicast traffic during fast reroute |
CN101330448A (zh) * | 2007-06-21 | 2008-12-24 | 华为技术有限公司 | 一种通告链路状态信息及确定组播转发路径的方法及装置 |
CN101453385A (zh) * | 2007-11-30 | 2009-06-10 | 华为技术有限公司 | 一种故障通告的方法及设备 |
CN101521927A (zh) * | 2009-04-03 | 2009-09-02 | 中兴通讯股份有限公司 | 一种组播转发路径收敛的方法和系统 |
Also Published As
Publication number | Publication date |
---|---|
CN103036787B (zh) | 2016-09-28 |
CN103036787A (zh) | 2013-04-10 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2013053276A1 (zh) | 一种网络路由收敛处理方法和装置 | |
US10516478B2 (en) | Controller based path estimation and path provisioning using optical impairment data | |
US9838246B1 (en) | Micro-loop prevention using source packet routing | |
EP3188409B1 (en) | Oam mechanisms for evpn active-active services | |
US8537720B2 (en) | Aggregating data traffic from access domains | |
JP5484590B2 (ja) | 擬似ワイヤに基づいてサービストラヒックを処理するための方法、デバイスおよびシステム | |
US8630303B2 (en) | Preventing loops in networks operating different protocols to provide loop-free topology | |
WO2009056034A1 (fr) | Procédé, système et équipement pour établir une détection bfd pour un tunnel lsp | |
WO2008098451A1 (fr) | Procédé d'établissement de tunnel, dispositif de noeud de réseau et système de réseau | |
WO2008025299A1 (fr) | Procédé de calcul de chemin de racine dans une passerelle de chemin le plus court | |
CN109889350A (zh) | 一种用于sdn网络故障中切换路径的方法及装置 | |
WO2013174101A1 (zh) | 连通性检测方法、装置和系统 | |
WO2016106482A1 (zh) | 误码信息传递方法和网络设备及通信系统 | |
EP2689561A1 (en) | Directing traffic in an edge network element operable to perform layer 2 data forwarding and supporting any of various spanning tree protocols | |
WO2020135339A1 (zh) | 一种网络路径收敛的方法以及相关设备 | |
WO2019006704A1 (zh) | 一种路径计算的方法、装置和系统 | |
WO2012146097A1 (zh) | Vpls网络和以太环网的倒换方法及装置 | |
WO2012119372A1 (zh) | 一种报文处理方法、设备和系统 | |
CN102957620A (zh) | 一种trill网络中mac地址表项的管理方法和设备 | |
WO2019170085A1 (zh) | 一种故障诊断方法及其装置 | |
CN113366804A (zh) | 防止网络拓扑改变期间的微环路的方法和系统 | |
WO2014029287A1 (zh) | 隧道负荷分担方法及装置 | |
US11296980B2 (en) | Multicast transmissions management | |
WO2015135284A1 (zh) | 数据流转发的控制方法及系统、计算机存储介质 | |
WO2012103729A1 (zh) | 隧道配置方法及装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 12840029 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 12840029 Country of ref document: EP Kind code of ref document: A1 |