A SDN single-link fault recovery method based on segment routing

Abstract

The present invention discloses a SDN single-link fault recovery method based on segment routing, comprising: collecting topology information through a controller; running a BPF algorithm based on the collected topology information to determine a set of feasible backup paths for the protected link based on the endpoint of each affected data packet; running a BPU algorithm to determine post-failure congestion on the determined backup path set based on resource utilization collected by the controller, updating the backup path set based on the determination, and selecting a final backup path. By introducing segment routing and a novel fault recovery algorithm, the present invention consumes fewer fault recovery forwarding rules and has better post-failure load balancing capabilities, achieving rapid recovery from single-link faults and avoiding congestion after fault recovery.

Description

Technical Field

[0001] The present invention belongs to the field of network fault recovery, relates to an SDN single-link fault recovery technology, and specifically relates to an SDN single-link fault recovery method based on segment routing. Background Art

[0002] Segment routing (SR), proposed by the Internet Engineering Task Force (IETF) in 2013, represents a new network concept—application-driven networking—that differs from traditional IP / MPLS networks. The core of segment routing is the source routing mechanism, where the forwarding path for a data packet is determined at the source node and encapsulated in the packet header. Segment routing uses node-segment identifiers (Node-SIDs) and adjacency-segment identifiers (Adj-SIDs) to represent nodes and links in the network, respectively. The data forwarding path is represented by a sequence of segment identifiers (SIDs). During data forwarding, intermediate nodes only need to perform operations on the packet based on the outermost SID in the packet header. Segment routing naturally supports SDN network architectures, allowing the SDN controller to centrally calculate explicit paths and deliver them to the source node in the form of SIDs.

[0003] Existing fault recovery solutions are generally categorized as active and passive. Fast Reroute (FRR) is an active fault recovery solution that achieves rapid recovery after a fault by pre-calculating a backup path. Most existing FRR solutions can achieve fast fault recovery within 50 milliseconds. Some segment routing-based FRR technologies, such as TI-LFA (Topology Independent Loop-Free Alternate), are also being deployed in real-world networks. However, whether in traditional IP networks or SDNs, the criteria for evaluating an FRR technology are often based on fault protection coverage, fault recovery time, and even the number of forwarding rules, with little consideration given to the rational allocation of resources after a fault. Under high traffic loads, traffic forwarding along the shortest IGP path can lead to frequent link congestion. This non-fault-induced congestion can be effectively addressed by deploying load balancing solutions such as traffic engineering (TE) and traffic scheduling. Similarly, when key network components experience sudden failures, large amounts of traffic are rerouted. Inappropriate traffic allocation can also lead to frequent link congestion, known as post-fault congestion. Existing FRR solutions can basically effectively reduce the impact of network connectivity loss, but almost no solution considers the problem of congestion after a failure. Summary of the Invention

[0004] Purpose of the invention: In order to overcome the shortcomings of the existing technology, a segment routing-based SDN single-link fault recovery method is provided. By introducing segment routing and a new fault recovery algorithm, fewer fault recovery forwarding rules are consumed, and better post-fault load balancing capabilities are achieved, thereby achieving rapid recovery from single-link failures and congestion avoidance after fault recovery.

[0005] Technical solution: To achieve the above objectives, the present invention provides an SDN single-link fault recovery method based on segment routing, comprising the following steps:

[0006] S1: Collect topology information through the controller;

[0007] S2: Based on the collected topology information, run the BPF algorithm to determine a set of feasible backup paths for the protected links according to the destination of each affected packet;

[0008] S3: Run the BPU algorithm to determine the congestion of the backup path set determined in step S2 after the failure based on the resource utilization collected by the controller, update the backup path set based on the determination, and select the final backup path.

[0009] Furthermore, the execution of the BPF algorithm in step S2 includes the following steps:

[0010] A1: Calculate the P space for the protected link;

[0011] A2: Calculate the Q space based on the endpoints of the affected traffic and use the post-convergence path as the default backup path.

[0012] A3: Based on the P space and Q space, 0-SID, 1-SID, and 2-SIDs backup paths are calculated for the protected link and the traffic endpoint, respectively.

[0013] Furthermore, the backup paths in the backup path set in step S2 satisfy constraints a and b:

[0014] a) The selected backup path is acyclic;

[0015] b) At most 2 SIDs are pushed into the PLR.

[0016] Constraint a is set to avoid routing loops. For example, if the backup path passes through the faulty path again, a routing loop will occur if the packet reaches the faulty path and is then sent back to the backup path.

[0017] Furthermore, the 0-SID, 1-SID, and 2-SIDs backup paths in step A3 are S f，y，0 , Sf,y,1 , S f,y,2 , calculate S according to different combinations of P space and Q space f，y，0 , S f，y，1 , S f,y,2 , the specific calculation method is as follows:

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] Where PLR(f) represents the local recovery node of the failed link f, P f represents the P space of PLR(f), Q y A represents the Q space of the endpoint y of the affected traffic, f,y,2 Represents P f And the adjacent links and Q of the nodes y and the intersection of the adjacent links of the nodes, N f,y,2 Representative A f,y,2 The upstream node of the link.

[0024] Furthermore, the execution of the BPU algorithm in step S3 includes the following steps:

[0025] B1: The controller polls the backup paths of the protected link and determines post-failure congestion based on the current traffic matrix;

[0026] B2: If the network load after the failure exceeds the available bandwidth of the backup path, go to step B3; otherwise, continue to use the default backup path provided in step S2;

[0027] B3: Select a backup path with sufficient link capacity and the least number of SIDs from the backup path set for update.

[0028] Furthermore, the backup path in step B3 is updated in the following manner: a backup path with sufficient link capacity is selected from the set of available backup paths in the order of 2-SIDs, 1-SID, and 0-SID.

[0029] Beneficial effects: Compared with the prior art, the present invention has the following advantages:

[0030] 1. Based on LFA technology, the default backup path is the same as the post-convergence path, providing faster fault recovery capabilities.

[0031] 2. Use segment routing technology to guide the forwarding of affected data packets on the backup path, reduce forwarding rules, and reduce resource consumption.

[0032] 3. A heuristic scheme is proposed, which dynamically and adaptively selects a backup path to avoid congestion after a failure according to the network status, thereby achieving load balancing after a failure. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] Figure 1 is a flow chart of the method of the present invention;

[0034] Figure 2 This is a diagram of the architecture of the present invention;

[0035] Figure 3 Schematic diagram of P space and Q space;

[0036] Figure 4 Backup paths generated for different P-space and Q-space combinations. DETAILED DESCRIPTION

[0037] The present invention is further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not used to limit the scope of the present invention. After reading the present invention, modifications of various equivalent forms of the present invention made by those skilled in the art all fall within the scope defined by the claims attached to this application.

[0038] The present invention provides an SDN single link fault recovery method based on segment routing, such as Figure 1 As shown, it includes the following steps:

[0039] S1: Collect topology information through the controller;

[0040] S2: Based on the collected topology information, run the BPF algorithm to determine a set of feasible backup paths for the protected links according to the destination of each affected packet;

[0041] like Figure 3 As shown, some terms in LFAFRR are used in the BPF algorithm:

[0042] P-space (PLR, F): The set of nodes reachable from PLR (using post-convergence paths) without passing through the failed link F.

[0043] Q-space (D, F): The set of nodes that can reach the destination D (using the post-convergence path) without passing through the failed link F.

[0044] The BPF algorithm runs in the following steps:

[0045] A1: Calculate the P space for the protected link;

[0046] A2: Calculate the Q space based on the endpoints of the affected traffic and use the post-convergence path as the default backup path.

[0047] A3: Based on P space and Q space, 0-SID, 1-SID, and 2-SIDs backup paths are calculated for the protected link and the traffic endpoint, respectively.

[0048] The backup paths in the backup path set satisfy constraints a and b:

[0049] a) The selected backup path is acyclic;

[0050] b) At most 2 SIDs are pushed into the PLR.

[0051] Due to constraint b), each backup path can only push two or fewer SIDs. Therefore, the backup path segment list configuration has the following three cases:

[0052] 0-SID: The segment routing path segment list contains no SIDs other than the SID representing the destination node. The PLR ​​forwards packets to adjacent nodes, and traffic is forwarded along the shortest IGP path from the adjacent node to the data destination.

[0053] 1-SID: In this case, the backup path includes an intermediate node Node-SID in addition to the Node-SID representing the endpoint, such as [u;t]. u is the Node-SID of an intermediate node, which is used to bypass certain faulty units by forwarding traffic to u.

[0054] 2-SIDs: The backup path is compressed with one Node-SID and one Adj-SID, such as [u; (u, v); t], where (u, v) represents the adjacent link from node u to node t.

[0055] The 0-SID, 1-SID, and 2-SIDs backup paths are S f，y，0 , S f,y，1 , S f，y，2 , calculate S according to different combinations of P space and Q space f，y，0 , S f，y，1 , S f，y，2 , the specific calculation method is as follows:

[0056]

[0057]

[0058]

[0059]

[0060]

[0061] Where PLR(f) represents the local recovery node of the failed link f, P f represents the P space of PLR(f), Q y A represents the Q space of the endpoint y of the affected traffic, f,y,2 Represents P f And the adjacent links and Q of the nodes y and the intersection of the adjacent links of the nodes, N f，y，2 Representative A f，y，2 The upstream node of the link.

[0062] S3: Run the BPU algorithm to determine the congestion of the backup path set determined in step S2 after the failure based on the resource utilization collected by the controller, update the backup path set based on the determination, and select the final backup path.

[0063] The operation of the BPU algorithm includes the following steps:

[0064] B1: The controller polls the backup paths of the protected link and determines post-failure congestion based on the current traffic matrix;

[0065] B2: If the network load after the failure exceeds the available bandwidth of the backup path, go to step B3; otherwise, continue to use the default backup path provided in step S2;

[0066] B3: Select the backup path with sufficient link capacity and the least number of SIDs from the backup path set for update. The update method is: select the backup path with sufficient link capacity from the available backup path set in the order of 2-SIDs, 1-SID, and 0-SID.

[0067] In step S2 of this embodiment, all available backup paths can be screened out by utilizing P space and Q space, and the number of SIDs used in these backup paths can be limited to 2 or less. Figure 4 As can be seen from (a), when the neighbor node of the PLR ​​is in the Q space of the data packet destination, the PLR ​​only needs to forward the traffic directly to the neighbor node. At this time, the backup path is the 0-SID backup path, because the PLR ​​does not need to perform additional SID operations. The neighbor node forwards them along the IGP shortest path to node D and does not pass through the faulty link F. The 1-SID backup path is Figure 4(b) The path passing through the intermediate node A or intermediate node B. Node A and node B are nodes that are included in both Q space and P space. The number of such nodes is determined by the network topology. By pushing the Node-SID of node A or node B on top of the destination SID, the PLR ​​can make the traffic along the backup path go to the intermediate node A or B first and then to the destination. Such a backup path ensures that the data packet neither passes through the faulty unit nor causes a loop. When the P space and Q space do not intersect, you can find the link connecting the P space and Q space, use the Adj-SID of the link, and the Node-SID of the upstream node of the link. Figure 4 As shown in (c), the PLR ​​pushes the Node-SID of node A and the Adj-SID of link AB. This ensures that traffic arriving at node A is forwarded to node B via link AB, and then forwarded from link B to the destination. This backup path corresponds to a 2-SIDs backup path. By using these three backup paths, traffic can be quickly rerouted in the event of a failure.

[0068] The execution process of the BPF algorithm in this embodiment is shown in Algorithm 1. First, the P space P is calculated for the protected link. f (See line 2 in Algorithm 1), and then calculate the Q space Q according to the endpoints of the affected traffic y , and use the post-convergence path as the default backup path DS f，y (See lines 4 and 5 in Algorithm 1). Then, 0-SID, 1-SID, and 2-SIDs backup paths S are calculated for the protected link and the traffic endpoint respectively. f，y,0 、S f，y，1 、S f，y，2 (See lines 6-10 in Algorithm 1).

[0069] Algorithm 1 BPF algorithm

[0070] Input: network topology G(V, E), protected link set F

[0071] Output: SID set S of available backup paths

[0072]

[0073] The execution process of the BPU algorithm in step S3 of this embodiment is shown in Algorithm 2. The BPU algorithm runs after the BPF. The controller polls the backup paths of the protected link and determines post-failure congestion based on the current traffic matrix. If the post-failure network load exceeds the available bandwidth of the backup path, the controller selects a backup path with sufficient link capacity and the fewest SIDs from the backup path set for update. If no post-failure congestion is determined, the default backup path provided by the BPF is continued to be used.

[0074] Algorithm 2BPU algorithm

[0075] Input: network topology G(V, E), protected link set F, SID set S calculated by BPF, traffic matrix T m

[0076]

[0077] Algorithm 2 first calculates the resource utilization in the network after removing the affected traffic. It then selects a backup path from the 2-SIDs backup path, the 1-SID backup path, and the 0-SID backup path that meets bandwidth requirements, has the fewest SIDs, and the largest remaining bandwidth to minimize link congestion after a failure. After the selection is complete, the shortest backup path in the PLR ​​is updated: the link corresponding to the Adj-SID, and the shortest path from the downstream node of the Adj-SID to the destination. The minimum link bandwidth of these three paths is the available bandwidth of the 2-SIDs backup path. If the available bandwidth meets the traffic requirements, the 2-SIDs backup path is used (see lines 6-10 in Algorithm 2). The available bandwidth of the 1-SID path is the minimum bandwidth of the path from the PLR ​​to the node represented by the Node-SID and the path from the node represented by the Node-SID to the destination. If the available bandwidth meets the traffic requirements, the 1-SID backup path is used (see lines 12-15 in Algorithm 2). The available bandwidth of a 0-SID path is the minimum bandwidth of the direct link from the PLR ​​to the neighbor node and the shortest path from the neighbor node to the destination. If the available bandwidth meets the traffic demand, this 0-SID backup path is used (see lines 16-20 in Algorithm 2). Algorithm 2 sets the 0-SID backup path to the highest priority because, while using more SIDs makes it easier to select a backup path with greater available bandwidth, adding too many SIDs also means taking detours, which usually results in higher transmission delays.

[0078] In order to implement the above solution, this embodiment also provides a single link fault recovery model (LFA-CA for short) with congestion avoidance capability. LFA-CA controls traffic forwarding after a failure by calculating a backup path for the protected link in advance. Figure 2As shown, the LFA-CA model uses a segment routing controller framework, which is suitable for SDN network architectures and can be configured with multiple forwarding tables based on LFA-CA requirements. The segment routing controller implements a traffic engineering database (TED), which contains network topology and resource utilization information for path calculation. When a new flow forwarding needs to be established, the request is sent to the segment routing controller. The controller uses the segment routing path calculation module to calculate the forwarding path and encode it into a segment list, which is then distributed to the ingress node of the traffic via the OpenFlow protocol. The core modules of LFA-CA mainly consist of two core algorithms: the BPF and the BPU. The BPF algorithm runs during network initialization. It calculates a set of available backup paths based solely on collected topology information and sets a default backup path. The default backup path is a loop-free post-convergence path for the protected link, namely the LFA path. During network operation, the segment routing controller runs the BPU algorithm, detecting backup paths for the protected link based on collected resource utilization information. If it detects that post-failure congestion will occur if this backup path is used, it selects a congestion-avoiding backup path from the available backup path set and updates it.

Claims

1. A SDN single link fault recovery method based on segment routing, characterized in that: The steps include: S1: Collect topology information through the controller; S2: Based on the collected topology information, run the BPF algorithm to determine a set of feasible backup paths for the protected links according to the destination of each affected packet; S3: Run the BPU algorithm to determine the congestion of the backup path set determined in step S2 after the failure based on the resource utilization collected by the controller, update the backup path set based on the determination, and select the final backup path; The execution of the BPF algorithm in step S2 includes the following steps: A1: Calculate the P space for the protected link; P-space (PLR, F): the set of nodes reachable from PLR without passing through the faulty link F; A2: Calculate the Q space based on the endpoints of the affected traffic and use the post-convergence path as the default backup path. Q-space (D, F): the set of nodes that can reach destination D without passing through the failed link F; A3: Based on P space and Q space, 0-SID, 1-SID, and 2-SIDs backup paths are calculated for the protected link and the traffic endpoint, respectively. The backup paths in the backup path set in step S2 satisfy constraints a and b: a) The selected backup path is acyclic; b) At most 2 SIDs are pushed into the PLR; The execution of the BPU algorithm in step S3 includes the following steps: B1: The controller polls the backup paths of the protected link and determines post-failure congestion based on the current traffic matrix; B2: If the network load after the failure exceeds the available bandwidth of the backup path, go to step B3; otherwise, continue to use the default backup path provided in step S2; B3: Select a backup path with sufficient link capacity and the least number of SIDs from the backup path set for update.

2. The SDN single link fault recovery method based on segment routing according to claim 1, characterized in that: In step A3, the 0-SID, 1-SID, and 2-SIDs backup paths are S f,y,0 , S f,y,1 , S f,y,2 , calculate S according to different combinations of P space and Q space f,y,0 , S f,y,1 , S f,y,2 , the specific calculation method is as follows: Where PLR(f) represents the local recovery node of the failed link f, P f represents the P space of PLR(f), Q y A represents the Q space of the endpoint y of the affected traffic, f,y,2 Represents P f And the adjacent links and Q of the nodes y and the intersection of the adjacent links of the nodes, N f,y,2 Representative A f,y,2 The upstream node of the link.

3. The SDN single link fault recovery method based on segment routing according to claim 1, characterized in that: The backup path selection method in step B3 is: selecting a backup path with sufficient link capacity from the available backup path set in the order of 2-SIDs, 1-SID, and 0-SID.

4. The SDN single link fault recovery method based on segment routing according to claim 3, characterized in that: The backup path is updated in step B3 as follows: Update the three paths consisting of the backup shortest path in the PLR, the link corresponding to the Adj-SID, and the shortest path from the downstream node of the Adj-SID to the destination. The minimum link bandwidth of these three paths is the available bandwidth of this 2-SIDs backup path. If the available bandwidth meets the traffic demand, this 2-SIDs backup path is used. The available bandwidth of the 1-SID path is the minimum bandwidth of the path from the PLR ​​to the node represented by the Node-SID and the path from the node represented by the Node-SID to the destination. If the available bandwidth meets the traffic demand, this 1-SID backup path is used. The available bandwidth of a 0-SID path is the minimum bandwidth of the direct link from the PLR ​​to the neighbor node and the shortest path from the neighbor node to the destination. If the available bandwidth meets the traffic requirements, the 0-SID backup path is used.

Images