WO2008138255A1

WO2008138255A1 - Route process method, route processor and router

Info

Publication number: WO2008138255A1
Application number: PCT/CN2008/070898
Authority: WO
Inventors: Yuanbo Zhu; Kui Zhang
Original assignee: Huawei Technologies Co., Ltd.; Tsinghua University
Priority date: 2007-05-14
Filing date: 2008-05-07
Publication date: 2008-11-20
Also published as: CN101309201A; CN101309201B

Abstract

A route process method, router processor and router are provided. The route process method includes that: establishing and maintaining the connection between each node in the network, scheduling the task of each node; each node runs the link state route protocol in accordance with the scheduled task, and computes the route, obtains the computed result of the route; obtaining the route computed result of each node and aggregating the routes. The route processor includes that: the task scheduling module, for scheduling the task; the protocol running module, for running the link state route protocol in accordance with the scheduled task, and computing the route, obtaining the computed result of the route; the route aggregation module, for aggregating the route computed result. The router includes a plurality of router processors.

Description

Route processing method, route processor and router

This application claims priority to Chinese Patent Application No. 200710102083.5, entitled "Route Processing Method, Route Processor and Router", filed on May 14, 2007, the entire contents of which are incorporated herein by reference. In the application.

Technical field

The present invention relates to computer network technologies, and in particular, to a route processing method, a route processor, and a router of a link state routing protocol.

Background technique

In the Internet, it is usually the router that forwards the Internet message, and the router that directs the forwarding decision is the forwarding information base. This forwarding information store stores routing information provided by different routing protocols. The routing protocol running by the router can be divided into the border gateway protocol between the autonomous systems and the internal gateway protocol within the autonomous system according to the application. Among them, an autonomous system refers to an internet network in which a management mechanism is implemented by a single management organization. The link state routing protocol is the most widely used internal gateway protocol, the system to the Intermediate System to Intermediate System Routing Protocol (ISS). The link state routing protocol is based on the shortest path first algorithm. The running process is: discovering and maintaining protocols through specific neighbors, discovering neighbors around the router, and spreading and collecting the autonomous system with its reliable flooding mechanism. The local link information of each router forms a so-called link state database in the link state routing protocol. In this way, each router running the link state routing protocol will soon obtain panoramic information about the entire network topology and agree. By applying the shortest path first algorithm to the link state database, all network reachability information of the autonomous system can be obtained, and a routing table of the link state routing protocol is formed, and these routes are loop-free. Then, the routing table of the link state routing protocol is sent to the global routing table of the router, and the routing information obtained from different routing protocols is synthesized by the global routing table, and the optimal route is selected according to certain rules to form a data plane guide. Forwarding information library forwarded by the text.

The traditional link state routing protocol operation mode is usually one or more routing protocols (including the link state routing protocol) as an operating system task, running on the routing processor of the router control plane, that is, the protocol running processing, Functions such as route calculation and management control are concentrated on a single route processor. In this mode of operation, the total number of neighbors that the link state routing protocol can support The number and routing capacity depend on the computing power of the routing processor and the capacity of the physical memory that is provisioned. With the expansion of the network and the flattened network design, the user usually hopes that the total number of neighbors and the routing capacity supported by the link state routing protocol can be improved with the increase of hardware resources such as the routing processor, and the link state is realized. Extension of the routing protocol, the traditional link state routing protocol operating mode cannot be extended by the link state routing protocol because it is concentrated on a single routing processor.

Summary of the invention

Embodiments of the present invention provide a route processing method, a route processor, and a router, which can implement extension of a link state routing protocol.

The embodiment of the invention provides a route processing method, including: establishing and maintaining a connection between nodes in a network, performing task scheduling on the node; the node runs a link state routing protocol according to the scheduled task, and performs route calculation to obtain a route. Calculating the result; obtaining the route calculation result of the node and performing route summarization.

The embodiment of the present invention provides a route processor, including: a task scheduling module, configured to perform task scheduling between routing processors; a protocol running module, configured to run a link state routing protocol according to the scheduled task, and perform routing The calculation is performed to obtain a route calculation result; the route summary module is configured to perform route summarization on the route calculation result.

The embodiment of the present invention provides a router, including a plurality of route processors, where the route processor includes: a task scheduling module, configured to perform task scheduling between routing processors; and a protocol running module, configured to perform tasks according to the scheduling The link state routing protocol is run, and the route calculation is performed to obtain the route calculation result. The route summary module is configured to perform route summarization on the route calculation result.

It can be seen from the above technical solutions that the traditional link state routing protocol operation mode is centralized on a single routing processor, and is limited by the processing capability of the routing processor itself, and the extension of the link state routing protocol cannot be implemented. Yes: Establish a connection between nodes in the network, and perform task scheduling on each node; each node runs a link state routing protocol according to the scheduled task, and performs route calculation to obtain a route calculation result; acquire the route of each node Calculate the results and perform route summarization. Because the operation mode of the embodiment of the present invention can dynamically distribute the task scheduling, protocol execution, and route summary operations of the link state routing protocol to each node, that is, each route processor, so that the resources of each node can be fully utilized, thereby realizing The scalability of the link state routing protocol.

DRAWINGS 1 is a distributed model diagram of a link state routing protocol according to an embodiment of the present invention;

2 is a flowchart of a routing processing method according to an embodiment of the present invention;

3 is a flowchart of performing task scheduling in a route processing method according to an embodiment of the present invention;

4 is a schematic diagram of a scheduling state machine according to an embodiment of the present invention;

5 is a schematic diagram of a state machine of a routing table according to an embodiment of the present invention;

6 is a schematic diagram of establishing a connection relationship between nodes according to an embodiment of the present invention;

Figure Ί is a flowchart of an active allocation task in a routing processing method according to an embodiment of the present invention;

8 is a flowchart of a passive allocation task in a routing processing method according to an embodiment of the present invention;

9 is a schematic diagram of an application network of a routing processing method according to an embodiment of the present invention;

10 is a schematic diagram of related information in an application network of a routing processing method according to an embodiment of the present invention; FIG. 11 is a schematic diagram of related information in an application network according to a routing processing method according to an embodiment of the present invention; 3 is a schematic diagram of related information in an application network; FIG. 13 is a schematic structural diagram of a routing processor according to an embodiment of the present invention;

14 is a schematic structural diagram of a task scheduling module of a routing processor according to an embodiment of the present invention;

FIG. 15 is a schematic structural diagram of a router according to an embodiment of the present invention.

detailed description

The traditional link state routing protocol operation mode is centralized on a single route processor, and is limited by the processing capability of the route processor itself, and the link state routing protocol cannot be extended. The route processing method provided by the embodiment of the present invention is The scalability of the link state routing protocol can be achieved by dynamically distributing the functions contained in the link state routing protocol to multiple routing processors.

The router's multi-router processor architecture typically consists of two parts: several nodes and the internal communication network between them. These nodes are usually organized into a distributed master-slave relationship, that is, one node is the master node and the other nodes are distributed slave nodes. Each node includes: a routing processor for running a router control plane task such as a routing protocol; an external interface through which the routing processor is connected to the forwarding plane to receive and send routing control messages; an internal interface, a routing processor Through this interface, messages are received and sent on the internal switched network, completing internal communication between the nodes.

For convenience of description, the routing processor is hereinafter referred to simply as a node.

The distributed model of the link state routing protocol in the embodiment of the present invention includes a task scheduling module, a protocol running module, and a routing table summary module. The task scheduling module is configured to discover and maintain each The connection relationship between the nodes, and the task allocation according to the processing capability of the node; the protocol running module is configured to run the link state routing protocol according to the task assigned by the scheduling module, perform route calculation, obtain a route calculation result, and generate a local routing table; The routing table summary module is configured to perform route summarization on the route calculation result, that is, it is used to collect the local routing table of each node.

The task scheduling algorithm of the embodiment of the present invention includes a scheduling state machine and a routing table state machine. The scheduling state machine divides the state of each node into a main scheduling state, a backup scheduling state, and a slave scheduling state according to a decision right assigned by the task. The functions of the scheduling state machine include: establishing and maintaining relationships between nodes of the distributed system; electing the primary scheduling node, the backup scheduling node, and the secondary scheduling node; performing task assignment. The routing table state machine divides the state of each node into a primary summary state, a backup summary state, and a summary state according to the execution right of the routing table summary task, and completes the summary operation of each local routing table.

The technical solutions of the embodiments of the present invention are described in detail below with reference to the accompanying drawings.

Under the scheduling of the scheduling state machine, the distributed implementation model can be mapped to a specific running model according to a specific scheduling policy. Figure 1 shows an example of a distributed model of a link state routing protocol. The tag 22 indicates a router, and the tag 33 indicates a node. The three circles from top to bottom in the node are the three components under the distributed model: the task scheduling module, the protocol running module, and the routing table summary module. By scheduling the scheduling of the state machine, there can be several allocation modes. An allocation mode, as shown in A, under the control of the scheduling state machine, completes the task assignment of protocol running and routing table summary work on multiple nodes, each node performs protocol running and routing table summarization, then each node The routing table summary module will only complete the local routing information summary and send it to the router's global routing table. The router's global routing table is stored in the router's other memory. Another allocation mode, as shown in B, allows the primary scheduling node to perform routing table summarization. Then the primary scheduling node is also the primary summary node. In this case, the primary summary node needs to summarize the routing information of each node before delivering it to the router. Global routing table. There is also an allocation mode as shown in C. Further, under the scheduling of the routing table state machine, the routing table aggregation operation is distributed to the non-primary scheduling node. The embodiment of the present invention will be mainly described in the mode shown by B, but is not limited thereto.

Referring to FIG. 2, it is a flowchart of a routing processing method according to an embodiment of the present invention, including:

Al, perform task scheduling;

Before describing the details of this step, let me introduce the process of scheduling the state machine. Figure 4 shows the scheduling state machine. The scheduling state machine is somewhat similar to the interface state machine with the OSPF protocol. Each node will execute this state machine. This state machine defines the following states: "inactive" state, in which the node does not run the link state routing protocol; in the "waiting" state, the node can run the link state routing protocol but does not enter the working mode; The mode, and the decision rights assigned by the task are divided into the "main scheduling", "backup scheduling" and "slave scheduling" states. The scheduling state machine performs the following functions driven by the trigger event: Establish and maintain the relationship between the nodes of the distributed system; perform election of the primary scheduling node, the backup scheduling node, and the secondary scheduling node; perform task assignment. The various arrows in Figure 4 mark the possible state transition process when the node is running the state machine. Depending on the nature of the triggering event, each node may change its state or may remain in its current state. But there will always be a consistent election result between all nodes.

As shown in Figure 4, if the node stops the routing protocol in any state, it changes to the "inactive" state. When the routing protocol is started, it changes from the "inactive" state to the "waiting" state, starts the timer, and starts. The heartbeat message is broadcast on the internal communication network. After the timer expires, the main scheduling node is elected. The rule is to elect the main scheduling node with the highest processing power value, followed by the backup scheduling node, and the rest are slave scheduling nodes. When the processing power values are the same, Comparing the internal network address values, the rule is generally that the election network address value is the largest as the primary scheduling node, followed by the backup scheduling node, and the rest are the secondary scheduling nodes. The tasks of the backup scheduling node and the slave scheduling node are allocated by the primary scheduling node.

Step A1 can be further divided into the following steps, please refer to FIG. 3, including:

Bl, establishing and maintaining relationships between nodes, electing a primary scheduling node, a backup scheduling node, and a secondary scheduling node;

When the network system is started, each node learns the existence of each other by periodically broadcasting heartbeat messages, and elects the primary scheduling node and the backup scheduling node according to its processing capability, and the rest are slave scheduling nodes. The processing capability information is carried in the heartbeat message. Thereafter, the scheduling node sends a heartbeat message only to the primary scheduling node, and the primary scheduling node broadcasts a heartbeat message on the internal network.

For example, the five nodes A, B, C, D, and E shown in Figure 6 change from the "inactive" state to the "waiting" state after receiving the configuration command to run OSPF, start the timer, and start internally. A heartbeat message is broadcast on the communication network. In the format of the heartbeat message, the type field is represented by 0 - 1 bits, the 2 - 3 bits represent the status field, 4 - 7 bits represent the address field, and 8 - 9 bits represent the processing capability field. Since the current state of each node is "waiting", the value of the state field of the heartbeat message is "waiting", and the "processing capability" field is the node. The remaining processing power of the previous node. By broadcasting the heartbeat message, each node obtains the neighbor's information and constitutes the following neighbor list.

The states of nodes A, B, C, D, and E in this table are waiting, and the processing capability values are 8, 6, 3, 2, and 3, respectively. After the timer expires, each node starts the election of the primary scheduling node. The rule is to find the primary scheduling node with the highest processing power value from the obtained neighbor set, followed by the backup scheduling node, and the rest are slave scheduling nodes. When the "processing power" value is the same, the internal network address value is compared. The rule is generally that the election network address value is the largest as the primary scheduling node, the second is the backup scheduling node, and the rest is the secondary scheduling node. Thus, node A is elected as the master scheduling node, B is elected as the backup scheduling node, and C, D, and E are slave scheduling nodes. Therefore, nodes A and B transition to the "master scheduling" state and the "backup scheduling" state, respectively, and nodes C, D, and E all transition to the "slave scheduling" state. Thereafter, A periodically broadcasts a heartbeat message to the internal network, whose status field is "main scheduling", and ^C, D, and E only send heartbeat messages to A, and the B-state field is "backup scheduling", C, D, and E. The status field is "From Schedule".

When a new node F joins, the state of the F node is "waiting", which starts the timer and starts broadcasting heartbeat messages to the internal network. Since B, C, D, and E are not "master scheduling" at this time, they will ignore the heartbeat message from F with the status "waiting", and the A state is "main scheduling", which will add F to the neighbor. List. On the other hand, since A periodically broadcasts a heartbeat message whose status field is "main scheduling", the periodic time is less than the timeout period of the timer. After receiving the heartbeat message from A, F finds that the primary scheduling node exists in the internal network. The "discovery master scheduling" event is then issued, node F transitions the state from "waiting" to "slave scheduling" and from then only sends a heartbeat message to A.

B2, the main scheduling node performs task assignment;

After the primary scheduling node is elected, the primary scheduling node starts the initial assignment of tasks. In this step, the task is assigned by the scheduling state machine.

According to the common application scenario of the link state routing protocol in the network, the scheduling state machine can adopt no The same split unit is used to perform task assignment. The different split units refer to one of a virtual private network instance such as a link state routing protocol, one of a multi-protocol process, or an area border router in a link state routing protocol. One of a plurality of regions, etc.

The link state routing protocol multi-VPN instance refers to the routing protocol instance running under each VPN. Each instance only performs the routing function in the VPN, and has nothing to do with the routing instances in other VPNs. Link state routing protocol Multi-protocol process means that each protocol process runs on a part of the same independent network under the same autonomous system, the so-called protocol routing domain. The routing function is completed in this independent sub-routing domain, and they are mutually unrelated. . A zone border router is a link state routing protocol that defines a router that is simultaneously connected to multiple zones within an autonomous system. The link state routing protocol divides the entire autonomous system into multiple regions in order to extend support for a large number of Autonomous Systems. These areas are simultaneously divided into backbone areas and non-backbone areas. The backbone area must be physically or logically connected, and each non-backbone area must be connected to this backbone area. The junction between the area and the area is the area border router, and the other routers are called the intra-area routers. The routers in the area complete the routing function in the area by running the shortest path algorithm, and the area border routers are responsible for flooding the routing information in all areas to other areas in the form of digest to help the routers in the area complete the inter-area routing function. Road State Routing Protocol This method of implementing inter-area routing borrows the horizontal splitting method in the distance vector routing protocol. Therefore, the area border router actually runs the shortest path algorithm on each of the connected areas, and the different areas are relatively independent networks.

The three kinds of irrelevance described above can be the task allocation algorithm of the scheduling state machine. When a plurality of areas of a regional border router are divided into units, each processing node runs a specific area, maintains neighbor relationships in the area, and synchronizes link state information of the area. When multiple VPN instances or multi-protocol processes are used to allocate units for distributed tasks, each node can run several link-state routing protocol processes, each of which is an independent link-state routing protocol. In a specific implementation, a configuration option can be provided to enable the user to make a final choice according to the specific application scenario of the protocol.

The following describes the OSPF protocol of the link state routing protocol as an example, and performs task allocation by using multiple area allocation units of the area border router.

After determining the relationship between the processing nodes and electing the primary scheduling node, the primary scheduling node in the "main scheduling" state triggers the "active task allocation" event when receiving the routing control message, and executes the main operation shown in FIG. The task assignment operation process; when the backup scheduling node under the "backup scheduling" or the slave scheduling node in the "scheduled" state receives the routing control message, the "passive task assignment" event is triggered, and the passive operation shown in FIG. 8 is performed. Task assignment process.

Please refer to Figure 7, the active task assignment operation process includes:

Cl, receiving a routing control message;

C2, determining whether the message processing task is responsible for the node, if yes, proceeding to step C3, if no, proceeding to step C4;

By looking up the relevant information in the local interface list, the master scheduling node can determine whether the packet processing task is responsible for the node.

C3, the routing control message will be processed;

C4, whether the message processing task has been assigned, if yes, go to step C6, if no, go to step C5;

The main scheduling node can find out whether the packet processing task has been allocated by searching for related information in the area list of the task allocation mapping table.

C5, designating a node to be responsible for the packet processing task, and proceeding to step C6;

C6, sending interface information to the corresponding node, informing it to join the task;

01. Modify the interface mapping table information.

Refer to Figure 8, for the passive task assignment operation process including:

Dl, receiving a routing control message;

D2. Determine whether the message processing task is responsible for the node. If yes, go to step D3. If no, go to step D4.

By backing up the scheduling node or from the scheduling node by looking up the relevant information in the local interface list, it can be determined whether the packet processing task is responsible for the node.

D3. The routing control message will be processed.

D4. Discard the routing control.

The following is specifically described in conjunction with an application network example shown in FIG. Figure 9 includes the router RT4 in the area (areaO), RT1, RT2 and area real), and also includes the area border router RT3, and additionally includes networks n1 to n6. The area border router RT3 is connected to two areas through three interfaces, ifO, ifl and if2, respectively, areaO and areal. Where ifO, ifl It is connected to n3 and n4 in areaO in turn, and if2 is connected to router RT4 in areal. RT3 consists of three nodes, A, B, and C (not shown). The current state is "main scheduling", "backup scheduling", and "slave scheduling", that is, eight, B, and C are the main scheduling. Nodes, backup scheduling nodes, and slave scheduling nodes.

One possible task assignment result is: C is responsible for the task of areaO, and B is responsible for the tasks of areal. The specific operations are as follows: Initially, the data of each node and the forwarding plane is as shown in FIG. The forwarding plane receives the OSPF packet from the interface i«). According to the interface mapping table (currently empty), it selects the default mode to broadcast the packet to the entire distributed system, so A, B and C both receive the interface i« ) OSPF packets sent with areaO information. B and C do not find the i«) information in their local interface list, and judge that the task is not responsible for themselves, so the OSPF packet is discarded. A first does not find the ifO information in the list of its own local interface, and judges that the task is not responsible for itself. Then, A checks whether the task has been allocated. It searches for the areaO in the area list of the task allocation mapping table (this time is empty). , but not found, and then judge that the task has not been assigned, then add areaO to the area list of the task allocation mapping table, and push out the node C with the largest remaining processing capacity in the node queue arranged in the order of "processing power". This results in the assignment result for the specified C responsible for the areaO task. A will perform the following operations: Fill in C corresponding to the areaO of the area list of the task allocation mapping table, set the port corresponding to interface ifO to C in the interface mapping table, send a message to C, and notify it to join the new task i«) . Thus, the result shown in Fig. 11 is obtained.

In this process, A is also responsible for periodically passing the interface mapping table to the forwarding plane. As shown in Figure 11. When receiving the OSPF packet, the forwarding plane sends the packet to the corresponding node in unicast mode according to the interface and the interface mapping table. Therefore, when the forwarding plane sends the message from the interface i«) to the control plane again, the message is sent directly to C, which reduces the processing overhead on A and B.

Similarly, when A receives an OSPF packet from interface if2, it has areal information. B and C still do the same processing as before. A also performs similar processing. Adds the area list of the task allocation mapping table to the areal, and pushes the node B with the largest remaining processing capacity in the node queue, and obtains the allocation result as the designated B responsible for the areal task. A performs the following operations for this: The node corresponding to the areal of the area list of the task allocation mapping table is filled in B, and the port corresponding to the interface if2 is set to B in the interface mapping table, and a message is sent to B to notify it to join the new task if2. When A receives an OSPF packet from the interface ifl, it carries the areaO information. B and C still do the same processing as before. A finds the areaO in the area list of the task allocation mapping table, and knows that the areaO task has been assigned to C, so directly set the port corresponding to the interface ifl to C in the interface mapping table, send a message to C, and notify it to join the new task ifl. . The result of Fig. 12 is thus obtained.

Thus, C and B are assigned to tasks areaO and areal respectively, and the interfaces they contain are i«), ifl, and if2. Messages from interfaces i«) and ifl are sent directly to node C, and messages from interface if2 are sent directly to node B.

In summary, the above description describes a method for task assignment in the embodiment of the present invention, which performs active task assignment in a "master scheduling" state, and performs passive task assignment in a "backup scheduling" or "slave scheduling" state. The task assignment algorithm is centralized to the main scheduling node to complete, and the corresponding task is directly sent from the forwarding plane to the corresponding processing node, so that each node will only receive the routing information in the area that it is responsible for, and then independently route. Calculation.

B3. Elect and broadcast the main summary node of the routing table.

The primary scheduling node selects the primary summary node, the backup summary node, and the secondary summary node according to the processing capabilities of each node collected by the heartbeat message, and sends it to each node through an internal message.

Here, the routing table state machine is used for scheduling. The scheduling algorithm of the routing table state machine depends on the algorithm of the scheduling state machine. The routing table state machine, as shown in Figure 5, is similar to the scheduling state machine, and divides each node into: "waiting" state, the node does not enter the working mode, that is, the related operations of the routing table summary are not required. In the working mode, the decision status of the task assignment is divided into the "main summary", "backup summary" and "from summary" status. If the node stops the routing protocol in any state, it transitions to the "inactive" state. When the routing protocol is started, it changes from the "inactive" state to the "waiting" state, starts the timer, and starts broadcasting on the internal communication network. Heartbeat message. After the timer expires, the route summary relationship is determined.

The routing summary relationship determining process refers to one of the three states in which the routing state of each node transitions from the "waiting" state to the working mode, that is, the initialization process of the routing table state machine is generally synchronized with the scheduling state machine initialization process. During the election of the main scheduling node of the scheduling state machine, each node transitions from the "waiting" state to one of the three states of scheduling, as the scheduling state machine "timer timeout" event triggering the operation and "discovering the main scheduling" The event is also a trigger event for the routing table state machine. Correspondingly, the node in the scheduling state machine that transitions from the "waiting" state to the "master scheduling" state, the routing table state machine will also be The "Wait" state transitions to the "Main Summary" state. That is to say, the initialization of the routing table state machine is the same as the initialization of the scheduling state machine. The node that performs the primary scheduling by default is also responsible for the routing table summary, that is, the primary scheduling node is also the primary summary node.

A2. Each node runs a link state routing protocol, performs route calculation, and obtains a local routing table. Each node runs a link state routing protocol locally according to the assigned task, discovers and maintains the neighbor relationship, and synchronizes the link state with each neighbor. Information, applying the shortest path first algorithm for routing calculation, and obtaining a local routing table.

A3. Perform route summarization.

Each node performs route summarization according to the route summary relationship determined by the routing table state machine.

Generally, each slave summary node routes the route to the primary summary node, and the primary summary node performs summary of the routing table to obtain the summary result, and delivers a copy to the backup summary node.

When the scheduling state machine is executed in the multi-zone or link state routing protocol multi-process mode of the link state routing protocol area border router, since the routing tables of the respective processes are all in the same range, each local routing table needs to be A global routing table that aggregates to form a complete routing protocol. Each of the slave routing nodes sends a local routing table to the primary summary node, and the primary summary node aggregates all the routing calculation results to form a global routing table of the routing protocol, and delivers the routing table to the global routing table of the router.

Although the default primary scheduling node is the primary summary node, which is responsible for the summary of the routing table, under the routing table state machine scheduling, the primary summary node may also be the same node as the primary scheduling node, that is, the task scheduling may be performed at one node. The routing table summary can be run on a different node.

It should also be noted that if the link state routing protocol multiple instance is used as the scheduling unit, since each protocol instance is valid in the respective VPN scope, each node can independently complete the summary of the respective routing tables, which is equivalent to Each node is a master summary state.

The summary of the routing table may also consider some specific requirements of the routing protocol. For example, when the OSPF area is a distributed execution unit, support for the OSPF virtual link needs to be considered. For the OSPF trunk area configured with the virtual link, the Link State Advertisement (LSA) of the Type 3 and Type 4 link status needs to update the route calculation result of the backbone area. Therefore, the two types of LSAs can be calculated locally before. A part of the local routing table is sent together to the main summary node for aggregation according to the protocol rules, and the specific process depends on the implementation of the routing protocol.

The summary of each local routing table will form a global routing table of the routing protocol, and the link state routing at this time The protocol needs to be sent to the global routing table of the router. The global routing table of the router integrates the routing information obtained from different routing protocols, and selects the optimal route according to certain rules to form a forwarding information base for the data plane to guide the packet forwarding. And sent to the forwarding plane.

The foregoing describes the routing processing method in the embodiment of the present invention. Correspondingly, the embodiment of the present invention provides a routing processor.

Please refer to FIG. 13, which is a schematic structural diagram of a routing processor according to an embodiment of the present invention. The route processor includes a task scheduling module 100, a protocol execution module 200, and a route summary module 300.

The task scheduling module 100 is configured to discover and maintain a connection relationship with other routing processors, and perform task allocation according to processing capabilities of each routing processor.

The protocol running module 200 is configured to run a basic link state routing protocol according to the assigned task, perform route calculation, and obtain a route calculation result, that is, a local routing table.

The route summary module 300 is configured to collect a local routing table obtained by routing calculation by each route processor, to obtain a global routing table of the complete routing protocol.

Referring to Figure 14, the task scheduling module 100 of the routing processor further includes an information unit 1001, a scheduling election unit 1002, a route summary election unit 1003, and a task assignment unit 1004.

The information unit 1001 is configured to store the routing processor's own processing capability value information and obtain processing capability value information of other routing processors; each routing processor learns the existence of each other by periodically broadcasting a heartbeat message, and the processing capability value information is in the heartbeat message. Carrying, the information unit 1001 obtains the processing capability value information of the other route processors through the heartbeat message.

The scheduling election unit 1002 is configured to elect a routing processor for performing task allocation according to the self-routing processor processing capability value information stored by the information unit 1001 and the acquired processing capability value information of other routing processors. The election rule is to find the main scheduling route processor with the highest processing capacity value for task assignment from the obtained neighbor set, followed by the backup scheduling route processor, and the rest is the slave routing processor. When the processing power values are the same, the internal network address values are compared. The rule is generally that the election network address value is the largest as the primary scheduling route processor, followed by the backup scheduling route processor, and the rest is the slave scheduling route processor.

The route summary selection unit 1003 is configured to elect, according to the self-processing capability value information stored by the information unit 1001 and the obtained processing capability value information of other route processors, a route processor that performs route summarization. The election rule is generally the primary summary route for routing summaries with the largest processing power value. The processor, followed by the backup summary route processor, and the rest is the summary route processor. Generally, the primary scheduling route processor elected by the scheduling election unit 1002 and the primary summary routing processor elected by the route summary election unit 1003 are the same routing processor, but the primary summary routing processor elected by the route summary election unit 1003 may also Not the same route processor as the main dispatch route processor.

The task allocating unit 1004 is configured to perform task allocation when the routing processor is used as the main scheduling route processor, and the task can be assigned to other routing processors.

The task scheduling module 100 of the routing processor further includes: a first timer unit 1005 and a scheduling state machine unit 1006.

The first timer unit 1005 is configured to set timing. The scheduling state machine unit 1006 is configured to divide the task state of the routing processor, and instruct the routing processor to perform different operations. When the routing processor enters a waiting state and starts timing at the first timer unit 1005, the scheduling state machine unit is scheduled. The notification information unit 1001 acquires the processing capability value information of the other routing processors. After the timing set by the first timer unit 1005 times out, the scheduling state machine unit 1006 triggers the scheduling election unit 1002 to perform the election, and instructs the routing processor to be elected as When the task is assigned, the user enters the main scheduling state, and is not elected to enter the backup scheduling state or the slave scheduling state when performing task assignment.

The task scheduling module 100 of the routing processor further includes: a second timer unit 1007 and a routing table state machine unit 1008.

The second timer unit 1007 is configured to set the timing. The routing table state machine unit 1008 is configured to divide the route summary state of the route processor, and instruct the route processor to perform different operations. When the route processor enters the waiting state and starts timing at the second timer unit 1007, the routing table The state machine unit 1008 notifies the information unit 1001 to acquire the processing capability value information of the other route processors. When the timing set by the second timer unit 1007 times out, the routing table state machine unit 1008 triggers the route summary election unit 1003 to perform the election and indicates the route. The processor is elected to enter the primary summary state when performing route summarization, and is not elected to perform backup summary status or summary status when performing route summarization.

The task assignment unit 1004 in the task scheduling module 100 of the route processor further includes: a determination unit 10041 and a processing unit 10042. The determining unit 10041 is configured to determine, according to the interface list information, whether the packet processing task belongs to the local routing processor, and the processing unit 10042 is configured to: when the determining unit 10041 determines that the routing processor is not responsible for the routing processor, The routing processor responsible for the packet processing task and modifying the interface mapping table information. The embodiment of the invention further provides a router. Please refer to FIG. 15, which is a schematic structural diagram of a router according to an embodiment of the present invention.

In Figure 15, the router contains three routing processors 60, 70, and 80, but is not limited thereto. The router may contain multiple routing processors. The routing processors of the router are the same as the routing processors described in FIG. 13, and include a task scheduling module 100, a protocol running module 200, and a routing summary module 300.

The task scheduling module 100 is configured to discover and maintain a connection relationship with other routing processors, and perform task allocation according to processing capability values of the routing processors. The protocol running module 200 is configured to run a basic link state routing protocol according to the assigned task, perform route calculation, and obtain a local routing table. The route summary module 300 is configured to perform route summarization on the route calculation result, and may collect a local routing table obtained by routing calculation by each route processor, thereby forming a global routing table of the complete routing protocol. The task scheduling module 100 of each routing processor further includes an information unit 1001, a scheduling election unit 1002, a route summary election unit 1003, a task allocation unit 1004, a first timer unit 1005, a scheduling state machine unit 1006, and a second timer unit 1007. And routing table state machine unit 1008. The task assignment unit 1004 in the task scheduling module 100 further includes a decision unit 10041 and a processing unit 10042. The functions of these units are the same as those described above and will not be described in detail here.

In summary, the link state routing protocol operating mode in the prior art is concentrated on a single routing processor, and is limited by the processing capability of the routing processor itself, and cannot implement the extension of the link state routing protocol, and the implementation of the present invention The example solution is: establishing a connection between nodes in the network, performing task scheduling on each node; each node runs a link state routing protocol according to the scheduled task, and performs route calculation to obtain a route calculation result; The routing calculation results are summarized and route summarization is performed. Because the operation mode of the embodiment of the present invention can dynamically distribute the task scheduling, protocol execution, and route summary operations of the link state routing protocol to each node, that is, each route processor, so that the resources of each node can be fully utilized, thereby realizing An extension of the link state routing protocol.

Further, the three functions of the link state routing protocol are dynamically distributed to each node, which can have greater fault tolerance. If a node of a distributed system fails, or a node's protocol runs during a large oscillation, it can isolate the impact on its nodes.

The foregoing provides a route processing method, a route processor, and a router provided by an embodiment of the present invention. It is to be noted that the description of the above embodiments is only for helping to understand the method and the core idea of the embodiments of the present invention. Meanwhile, for the person skilled in the art, according to the idea of the embodiment of the present invention, in the specific implementation manner and application scope There are variations, and the description is not to be construed as limiting the invention.

Claims

Rights request

A route processing method, comprising:

Establish and maintain connections between nodes in the network, and schedule tasks for nodes;

The node runs the link state routing protocol according to the scheduled task, and performs route calculation to obtain a route calculation result.

Obtain the route calculation result of the node and perform route summarization.

The route processing method according to claim 1, wherein the task scheduling of the pair of nodes is specifically:

Determining a node for performing task assignment and a node for performing route summarization according to processing capability value information of each node;

The nodes that perform the task assignment are assigned tasks for each node.

The method for processing a route according to claim 2, wherein: determining, according to the processing capability value information of each node, a node that performs task assignment and a node that performs route summarization are: acquiring a mobile node that is broadcasted in the network. The heartbeat message of the capability value information is based on the processing capability value information, and the node with the highest processing capability value is used as a node for performing task assignment and a node for performing route summarization.

The route processing method according to claim 2, wherein: the node that determines the task assignment according to the processing capability value information of each node is controlled by the scheduling state machine, and the control process is specifically:

The node that enters the waiting state broadcasts a heartbeat message carrying the node state information and the processing capability value information after the timer is started. After the timer expires, the election process of determining the node for performing task assignment is triggered, and the node with the largest processing capability value is elected. As a node for performing task assignment, indicating that the node with the highest processing capability value enters a primary scheduling state; indicating that the node with the second processing capability value is the backup scheduling state, and the remaining nodes are the slave scheduling states; The heartbeat message is sent only from the node in the scheduling state to the node in the primary scheduling state.

The routing processing method according to claim 2, wherein: the node that performs route summarization according to the processing capability value information of each node is controlled by a routing table state machine, and the control process is:

Instructing the node entering the waiting state to broadcast the carrying node status information and processing after starting the timer The heartbeat message of the capability value information, after the timer expires, triggers the election process of the route summary node, elects the node with the highest processing capability value as the node that performs route summarization, and indicates that the node with the highest processing capability value enters the main scheduling state. And indicating that the node with the second processing capability value is in the backup scheduling state, and the remaining nodes are sent from the summary state, and the node in the summary state only sends the heartbeat message to the node in the primary summary state.

The routing processing method according to claim 2, wherein the determining the task of assigning each node by the node performing task assignment is specifically:

Determining that the node performing task assignment allocates tasks of each node according to different split units, one of the multiple virtual private network instances of the link state routing protocol, one of the multi-protocol processes, or one of the area border router connections. One of the multiple regions.

The routing processing method according to claim 2 or 6, wherein the task of assigning each node by the node that determines the task assignment is specifically:

After the determined node that receives the task receives the message, the node list information is queried, and if it is determined that the message does not belong to the node, the node responsible for the message processing task is specified, and the interface mapping table is modified. information.

The route processing method according to claim 2, wherein the obtaining the route calculation result of the node and performing route summarization is specifically:

Each node sends a local routing table that performs routing calculation results to the determined node for performing route summarization, and the node that performs route summarization performs route summing on the received local routing table to obtain a global routing table.

9. A routing processor, comprising:

a task scheduling module, configured to perform task scheduling between routing processors;

a protocol running module, configured to run a link state routing protocol according to the scheduled task, and perform route calculation to obtain a route calculation result;

A route summary module, configured to perform route summarization on route calculation results.

The route processor according to claim 9, wherein the task scheduling module further comprises:

An information unit, configured to store routing processor own processing capability information and obtain processing capability information of other routing processors; a scheduling election unit, configured to elect a routing processor for performing task allocation according to the routing processor's own processing capability information stored by the information unit and the obtained processing capability information of the other routing processor; and a route summary selecting unit for storing according to the information unit The node processing capability information and the acquired processing capability information of the other routing processors elect a routing processor that performs route summarization;

A task allocation unit is configured to perform an allocation task when the task scheduling module is in a primary scheduling state.

The route processor according to claim 10, wherein the task scheduling module further comprises:

a first timer unit, configured to set a timing;

a scheduling state machine unit, configured to divide a task state of the routing processor, and instruct the routing processor to perform different operations, and when the routing processor enters a waiting state and after the first timer unit starts timing, the scheduling state machine unit notifies the information The unit obtains the processing capability information of the other routing processor. After the timing set by the first timer unit expires, the scheduling state machine unit triggers the scheduling election unit to perform the election, and indicates that the routing processor is elected to perform the task allocation and enters the main scheduling state. , not elected to enter the backup scheduling state or slave scheduling state when performing task assignment.

a second timer unit, configured to set a timing;

a routing table state machine unit, configured to divide a route summary state of the route processor, and instruct the route processor to perform different operations, and when the route processor enters a wait state and starts timing at the second timer unit, the routing table state machine The unit notification information unit acquires the processing capability information of the other route processor. After the timing set by the second timer unit expires, the routing table state machine unit triggers the route summary election unit to perform the election, and instructs the route processor to be elected as the route summary. When entering the main summary state, it is not elected to enter the backup summary state or from the summary state when performing route summarization.

The route processor according to claim 10, wherein the task allocation unit further comprises:

a determining unit, configured to determine, according to the interface list information, whether the message processing task belongs to the routing processor according to the interface list information;

The processing unit is configured to: when the determining unit determines that it is not responsible for the routing processor, specify a routing processor responsible for the packet processing task, and modify the interface mapping table information.

A router, comprising: a plurality of routing processors, the routing processor comprising: