In-band telemetry path planning algorithm based on segment routing

By using optimized Euler path planning and overlapping path elimination algorithm in segment routing environment, the problem of path redundancy and insufficient coverage in in-band telemetry technology is solved, and more uniform flow distribution and efficient telemetry path planning are achieved, improving network reliability and coverage.

WO2024229893A1PCT designated stage expired Publication Date: 2024-11-14SOUTHEAST UNIV

Patent Information

Application Number
PCT/CN2023/095962
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-05-10
Filing Date
2023-05-24
Publication Date
2024-11-14

AI Technical Summary

Technical Problem

The existing in-band network telemetry technology cannot independently determine the monitoring path, resulting in concentrated measurement path areas and redundancy problems, which fails to effectively improve the coverage of telemetry.

Method used

The in-band telemetry path planning algorithm based on segment routing is adopted. By optimizing the Euler path planning algorithm and overlapping path elimination algorithm, the active flow in the network is judged and sub-graphs are generated, reducing redundant measurements and improving coverage.

Benefits of technology

It realizes reducing redundant measurement of links during flow-by telemetry, improving the coverage of telemetry, reducing the redundancy of path planning, and improving the reliability and flexibility of the network.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023095962_14112024_PF_FP_ABST
    Figure CN2023095962_14112024_PF_FP_ABST
Patent Text Reader

Abstract

Disclosed in the present invention is an in-band telemetry path planning algorithm based on segment routing. After starting points and ending points of flows are determined, according to historical telemetry data such as paths through which the remaining flows pass, the network congestion degree, and other information, the flows are guided to be distributed in a network more uniformly. According to the present invention, a network topology is modeled into an undirected connected graph, the graph is redivided according to the activity level of each flow, and then an auxiliary node is added to eliminate repeated links, thereby completing adaptive path planning for a network topology structure. In addition, the feasibility of the invention is verified by means of cases such as a randomly generated special topology structure and real data center topology.
Need to check novelty before this filing date? Find Prior Art

Description

An in-band telemetry path planning algorithm based on segment routing Technical Field

[0001] The present invention belongs to the intersection field of network security and network measurement, and in particular relates to an algorithm for performing path planning for in-band telemetry in a segment routing environment. Background Art

[0002] Segment Routing (SR) is a source routing protocol whose core concept is to segment the network packet forwarding path into distinct segments and insert segmentation information into the packet at the path starting point to guide packet forwarding. Based on the concept of SDN, segment routing forms a path-connected network architecture that supports the multi-layered programmability requirements of future networks and can meet the connectivity needs of 5G ultra-large connections and slicing scenarios. Its purpose is to simplify the network and improve controllability. It supports multiple forwarding planes, such as MPLS and IPv6, and offers advantages in traffic engineering, network programming, and application-driven development.

[0003] In-band network telemetry is a hybrid measurement technology that enables the collection, transmission, organization, and reporting of network status by embedding telemetry instructions and metadata within data packets. In an in-band network telemetry architecture, switching devices forward and process packets carrying telemetry instructions. As telemetry packets pass through the device, these instructions inform the telemetry-enabled network device of the network status information to collect and record. Its advantage is that it requires no additional control plane intervention and can monitor network topology, performance, and traffic in real time, significantly assisting in scenarios such as network fault location, congestion control, and path decision making. Various telemetry solutions have been developed in academia, but as a low-level primitive, INT only defines how to extract internal device status, such as ingress rate and egress traffic, and cannot proactively determine which path to monitor. For example, when the INT header is embedded in an IP packet, the telemetry path is passively determined by its destination IP address and each switch's routing table, completely outside the control of the INT server. Path planning also fails to consider factors such as telemetry coverage. As a result, when in-band measurements are performed along a flow, the measured path is relatively concentrated and suffers from significant redundancy.

[0004] In summary, the present invention uses the in-band telemetry technology of segment routing and performs path planning by optimizing the Euler path planning algorithm and the overlapping path elimination algorithm.

[0005] Summary of the Invention

[0006] To address these issues, this paper proposes an in-band telemetry path planning algorithm based on segment routing. This algorithm adaptively plans paths to reduce redundant measurements. After determining the start and end points of a flow, it combines historical INT data, such as the paths of other flows and network congestion levels, to guide a more even distribution of flows within the network. During in-band telemetry, this reduces redundant link measurements and improves telemetry coverage.

[0007] To achieve the objectives of the present invention, the technical solution is as follows: an in-band telemetry path planning algorithm based on segment routing determines active flows in the network and generates a subgraph G′ based on their original paths. An optimized Euler path algorithm is used to generate a path planning result set E′. Finally, an overlapping path elimination algorithm is used to remove redundant paths. The algorithm includes the following steps:

[0008] (1) Modeling network topology and network flow. Model the network topology as an undirected connected graph G, and model the flow as a six-tuple;

[0009] (2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0010] (3) Under the condition of determining the corresponding relationship between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0011] (4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning results.

[0012] As an improvement of the present invention, the specific method of step (1) is as follows:

[0013] (1.1) The network topology is modeled as an undirected connected graph G = (V, E), where each device in the network is abstracted as a vertex in the graph and each link is abstracted as an edge in the graph;

[0014] (1.2) The flow in the network is abstracted into a six-tuple f = (tuple, r, flag, t s ,t e ,path), where tuple is a five-tuple of data packets, namely source IP address, destination IP address, source port, destination port and transport layer protocol. Data packets with the same five-tuple constitute a flow;

[0015] (1.3) flag is the flag indicating whether the flow is active. 1 indicates that the flow is active and can be included in the measurement range. 0 indicates that the flow is inactive and is not included in the measurement range.s is the start time of the flow, t e is the end time of the flow, then t e -t s The duration of the flow can be calculated; path represents the path that the flow passes through. In the segment routing environment, path can be expressed as path(f)=(SID[n],…,SID[2],SID[i])=(v1,v2,…,v n ), where v1∈V,(v1,v1)∈E;

[0016] (1.4) Time t e -t s All active flows constitute the set F = {f1,f2,…,f n}, for any of the flows, there is an original forwarding path, i.e.

[0017] Through this step, the network topology is abstractly modeled into a graph, and the path planning problem is abstracted into a graph theory problem, which is convenient for implementation using algorithms.

[0018] As an improvement of the present invention, the design method for active flow judgment and connected subgraph generation in step (2) is as follows:

[0019] (2.1) Set the active flow threshold;

[0020] (2.2) Based on historical INT data, determine the activity level of the flow. If it is an active flow, set its flag to 1; if it is an inactive flow, set its flag to 0;

[0021] (2.3) At the cost of generating duplicate paths, transform the original paths of all active flows into a connected subgraph G′ of graph G;

[0022] (2.4) Determine the correspondence between the starting point and the end point of each flow in the connected subgraph G′.

[0023] Through this step, according to the flow activity level, the path of the existing flow is converted into a connected graph G into a connected subgraph G′ of graph G, and the starting point and end point of each flow are fixed.

[0024] As an improvement of the present invention, the specific method of the path planning algorithm in step (3) is as follows:

[0025] (3.1) Find the path with the largest coverage between the odd points in the graph G′;

[0026] (3.2) Remove the planned paths from the path set;

[0027] (3.3) Find a path from the starting point that is as non-repeating as possible;

[0028] (3.4) Generate the path planning result set E′.

[0029] Through this step, a planning result set with duplicate paths is preliminarily generated, so that the overlapping path elimination algorithm can be used to remove redundant paths.

[0030] As an improvement of the present invention, the specific method of the overlapping path elimination algorithm in step (4) is as follows:

[0031] (4.1) Count the overlapping paths in the path planning set E′;

[0032] (4.2) Select adjacent nodes in {GG′}, introduce paths outside of G′, and eliminate overlapping paths as much as possible. When selecting nodes, due to practical limitations, only nodes with adjacent links can be selected; nodes cannot be fabricated out of thin air;

[0033] (4.3) Use auxiliary nodes to construct new paths and reduce path planning redundancy;

[0034] (4.4) Generate the path result set R after removing redundancy.

[0035] Through this step, a path planning result set based on segment routing is generated. Compared with traditional methods, due to the controllability of segment routing, flow path planning with lower redundancy can be obtained, which improves the effectiveness of telemetry data for subsequent analysis.

[0036] Compared to existing technologies, the present invention offers the following advantages: This solution incorporates traditional in-band network telemetry (INT) into segment routing technology through a segment routing-based in-band telemetry path planning algorithm. The path is written into the SR extension header in SID format before packet forwarding, making it unaffected by routing during forwarding. Therefore, this work fully leverages this feature. After determining the start and end points of a flow, it combines historical INT data, such as the paths of other flows and network congestion levels, to guide a more even distribution of flows within the network. This reduces redundant link measurements during in-band telemetry and improves telemetry coverage. Furthermore, the unique nature of segment routing improves network reliability and flexibility, supports multiple candidate path generation methods, and implements path selection and load balancing based on priority and weighting. Furthermore, forwarding based on native IPv6 or MPLS offers excellent compatibility and scalability, without changing the original packet encapsulation structure. It can coexist with standard IPv6 or MPLS equipment and can be integrated into data center networks and even user terminals, promoting cloud-network convergence. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 is a flow chart of in-band telemetry path planning based on segment routing;

[0038] Figure 2 is a schematic diagram of the Euler path algorithm;

[0039] Figure 3 is a schematic diagram of single flow path overlap;

[0040] FIG4 is a schematic diagram of multi-flow path overlap;

[0041] Figure 5 is a schematic diagram of an auxiliary node;

[0042] Figure 6 is a diagram of the NSFNet topology;

[0043] Figure 7 is a diagram of the GBN topology;

[0044] Figure 8 is a diagram of the GEANT topology;

[0045] Figure 9 is a topological diagram of Germany50. DETAILED DESCRIPTION

[0046] In order to deepen the knowledge and understanding of the present invention, the present invention is described in detail below with reference to the accompanying drawings.

[0047] Example: Referring to Figures 1-9 , to achieve the objectives of the present invention, the technical solution is as follows: an in-band telemetry path planning algorithm based on segment routing determines active flows in the network and generates a subgraph G′ based on their original paths. An optimized Euler path algorithm is used to generate a path planning result set E′, and finally an overlapping path elimination algorithm is used to remove redundant paths. The algorithm includes the following steps:

[0048] (1) Modeling network topology and network flow. Model the network topology as an undirected connected graph G, and model the flow as a six-tuple;

[0049] (2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0050] (3) Under the condition of determining the corresponding relationship between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0051] (4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0052] The specific method of step (1) is as follows:

[0053] (1.1) The network topology is modeled as an undirected connected graph G = (V, E), where each device in the network is abstracted as a vertex in the graph and each link is abstracted as an edge in the graph;

[0054] (1.2) The flow in the network is abstracted into a six-tuple f = (tuple, r, flag, t s ,t e ,path), where tuple is a five-tuple of data packets, namely source IP address, destination IP address, source port, destination port and transport layer protocol. Data packets with the same five-tuple constitute a flow;

[0055] (1.3) flag is the flag indicating whether the flow is active. 1 indicates that the flow is active and can be included in the measurement range. 0 indicates that the flow is inactive and is not included in the measurement range. s is the start time of the flow, t e is the end time of the flow, then t e -t s The duration of the flow can be calculated; path represents the path that the flow passes through. In the segment routing environment, path can be expressed as path(f)=(SID[n],…,SID[2],SID[i])=(v1,v2,…,v n ), where v1∈V,(v1,v1)∈E;

[0056] (1.4) Time t e -t s All active flows constitute the set F = {f1,f2,…,f n}, for any of the flows, there is an original forwarding path, i.e.

[0057] The design method for active flow judgment and connected subgraph generation in step (2) is as follows:

[0058] (2.1) Set the active flow threshold;

[0059] (2.2) Based on historical INT data, determine the activity level of the flow. If it is an active flow, set its flag to 1; if it is an inactive flow, set its flag to 0;

[0060] (2.3) At the cost of generating duplicate paths, transform the original paths of all active flows into a connected subgraph G′ of graph G;

[0061] (2.4) Determine the correspondence between the starting point and the end point of each flow in the connected subgraph G′.

[0062] The specific method of the path planning algorithm in step (3) is as follows:

[0063] (3.1) Initialize the path planning set E′, find the longest Euler path among the odd-degree nodes in the graph G′ and add it to the path planning set E′;

[0064] (3.2) Remove the planned paths in the path set E′ from the mapping relationship F′ between the flow start point and the end point;

[0065] (3.3) In the mapping relationship F′ between the flow start point and the end point, for each Startpoint, use it as the starting point and repeat the operations (3.1) to (3.2) to find a non-repeated path;

[0066] (3.4) Generate the path planning result set E′.

[0067] The specific method of the overlapping path elimination algorithm in step (4) is as follows:

[0068] (4.1) Initialize the path set R. In the path planning set E′, count the overlapping paths. If there are duplicate paths, add them to the overlapping path set T;

[0069] (4.2) Select adjacent nodes in {GG′} to construct the auxiliary node point. When selecting nodes, due to practical limitations, we cannot fabricate nodes out of thin air, so we can only select nodes that are adjacent to the link;

[0070] (4.3) Use the auxiliary nodes and the current overlapping paths to construct a new path, remove the overlapping path path in the overlapping path set T, and update the path set R;

[0071] (4.4) Repeat the operation in (4.3) to generate the path result set R after removing redundancy.

[0072] Specific implementation method:

[0073] As shown in Figure 1, it is a flow chart of in-band telemetry path planning based on segment routing.

[0074] First, several network flows are input. The present invention uses the following formula to determine the activity level of a certain flow:

[0075] In t e -t s During the time period, the total amount of data sent by the flow is (t e -t s )×r, if:

[0076] The flow is judged to be active and its flag is set to 1. i The value of is used to filter out the existing active flows, flow TH The threshold for determining whether the current flow is active is calculated as shown in the formula:

[0077] From the above formula, we can see that the setting of the threshold has a great relationship with the flow rate variance. (user-defined parameter value), flow TH The value of will tend to 0, that is, when the amount of data sent by all flows in a certain time slice is relatively uniform, the judgment of active flows can be relaxed; when the variance is large, the setting of the threshold is related to the rate deviation and the proportion of large flows to all active flows. That is, if the amount of data sent by the current flow in a certain time slice is much smaller than the average of the current active flows, then the flow is more likely to be judged as an inactive flow.

[0078] To further illustrate the segment-routing-based in-band telemetry path planning algorithm, the Euler path planning algorithm will be introduced. As shown in Figure 2, the generated paths strictly meet the requirement of non-overlapping paths. However, the start and end points of the generated paths may not meet the requirements of this study, and the requirement for an odd degree at the start point of the path is overly stringent. In a real-world environment, if node B can only serve as the end point of a flow, the generated paths may be unusable.

[0079] Next, as shown in Figures 3 and 4, there are two main situations in which links in the network topology are repeatedly measured: one is due to path planning problems, which causes the link to be passed by two or more telemetry paths, that is, multiple telemetry paths have common edges; the second is due to certain requirements of the flow, which causes the link to be passed by the flow twice or more times, that is, So that (v h ,v h+1 )==(v k ,v k+1 Regardless of the reason, every time a link is measured more than once, a node is added to the graph, and a new path is derived through the node. The idea behind adding auxiliary nodes is to eliminate overlapping paths and increase the telemetry range by adding new paths. It also transforms even-degree nodes into singular points, facilitating the execution of the path planning algorithm.

[0080] To further illustrate the segment-routing-based in-band telemetry path planning algorithm, we next introduce an overlapping path elimination algorithm. If there are duplicate paths in the generated path result set E′, adjacent nodes are selected from {GG′}, paths outside of G′ are introduced, and overlapping paths are eliminated as much as possible. Due to practical limitations, only nodes adjacent to the link can be selected when selecting nodes, and no new nodes can be fabricated. Figure 5 shows a schematic diagram of adding nodes. It can be seen that the original overlapping links can reduce redundancy and add new measurement paths after the introduction of new nodes and new paths.

[0081] The present invention is described below with reference to specific examples and the accompanying drawings. The invention comprises the following steps:

[0082] (1) Modeling network topology and network flow. Model the network topology as an undirected connected graph G, and model the flow as a six-tuple;

[0083] (2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0084] (3) Under the condition of determining the corresponding relationship between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0085] (4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning results.

[0086] To illustrate the working steps of an in-band telemetry path planning algorithm based on segment routing, five use cases are used to illustrate the working mode of the algorithm under different topology diagrams.

[0087] Specific embodiment 1:

[0088] If this algorithm uses a special network topology structure with randomly generated topology, the workflow of an in-band telemetry path planning algorithm based on segment routing is as follows.

[0089] Use Python to randomly generate a special network topology with 2k+1 odd-degree nodes. In the Euler path algorithm, for an undirected connected graph G = (V, E) with 2k odd-degree nodes, there are at least k different paths {p1, p2…, p k}, and E(G)=E(p1)∪E(p2)…∪E(p k ), that is, these non-overlapping paths will cover all edges in the graph. In a network topology with 2k+1 odd-degree nodes, the set of path plans that can cover all edges in the graph must have path overlap.

[0090] When using the segment routing-based in-band telemetry path planning algorithm to plan a path for this topology graph, the specific steps are as follows:

[0091] (1.1) Model a special network topology with 2k+1 odd-degree nodes that is randomly generated as an undirected connected graph G, and model the flow in the network as a six-tuple;

[0092] (1.2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0093] (1.3) Under the condition of determining the correspondence between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0094] (1.4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0095] Specific embodiment 2:

[0096] If the topology used in this algorithm is the NSFNet network topology diagram, the workflow of an in-band telemetry path planning algorithm based on segment routing is as follows.

[0097] As shown in Figure 6, the NSFNet network topology consists of 14 nodes and 18 edges, of which 8 are odd-degree nodes. Therefore, there are at least 4 non-overlapping paths that can cover all edges in the topology graph.

[0098] When using the segment routing-based in-band telemetry path planning algorithm to plan a path for this topology graph, the specific steps are as follows:

[0099] (1.1) Model the NSFNet network topology as an undirected connected graph G and model the flows in the network as six-tuples;

[0100] (1.2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0101] (1.3) Under the condition of determining the correspondence between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0102] (1.4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0103] Specific embodiment 3:

[0104] If this algorithm uses a special network topology structure with randomly generated topology, the workflow of an in-band telemetry path planning algorithm based on segment routing is as follows.

[0105] As shown in Figure 7, the GBN network topology consists of 17 nodes and 26 edges, of which 6 are odd-degree nodes. Therefore, there are at least 3 non-overlapping paths that can cover all edges in the topology graph.

[0106] When using the segment routing-based in-band telemetry path planning algorithm to plan a path for this topology graph, the specific steps are as follows:

[0107] (1.1) Model the GBN network topology as an undirected connected graph G, and model the flows in the network as six-tuples;

[0108] (1.2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0109] (1.3) Under the condition of determining the correspondence between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0110] (1.4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0111] Specific embodiment 4:

[0112] If this algorithm uses a special network topology structure with randomly generated topology, the workflow of an in-band telemetry path planning algorithm based on segment routing is as follows.

[0113] As shown in Figure 8, the GEANT network topology consists of 24 nodes and 37 edges, of which 8 are odd-degree nodes. Therefore, there are at least 4 non-overlapping paths that can cover all edges in the topology.

[0114] When using the segment routing-based in-band telemetry path planning algorithm to plan a path for this topology graph, the specific steps are as follows:

[0115] (1.1) Model the GEANT network topology as an undirected connected graph G and model the flows in the network as six-tuples;

[0116] (1.2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0117] (1.3) Under the condition of determining the correspondence between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0118] (1.4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0119] Specific embodiment 5:

[0120] If this algorithm uses a special network topology structure with randomly generated topology, the workflow of an in-band telemetry path planning algorithm based on segment routing is as follows.

[0121] As shown in Figure 9, the Germany50 network topology consists of 50 nodes and 87 edges, of which 26 are odd-degree nodes. Therefore, there are at least 13 non-overlapping paths that can cover all edges in the topology.

[0122] When using the segment routing-based in-band telemetry path planning algorithm to plan a path for this topology graph, the specific steps are as follows:

[0123] (1.1) Model a randomly generated special network topology with 2k odd nodes into an undirected connected graph G, and model the flow in the network into a six-tuple;

[0124] (1.2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into a connected subgraph G′;

[0125] (1.3) Under the condition of determining the correspondence between the starting point and the end point of each flow in the connected graph G′, the optimized Euler path algorithm is used to generate the path result set E′ in the connected graph G′ and plan a network telemetry path solution with a small number of overlaps;

[0126] (1.4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, reducing redundancy and generating the final path planning result set.

[0127] Experiment 1: Simulation test in a simulated environment

[0128] Purpose of the experiment:

[0129] Test the total number of links detected by the segment routing-based in-band telemetry path planning algorithm in randomly generated topology graphs and real data center topology graphs

[0130] Experimental steps:

[0131] 1. Set up an INT simulation environment in Mininet and use Python to randomly generate topology diagrams with different numbers of vertices and real data center network topology diagrams to determine the correspondence between the starting point and end point of each flow;

[0132] 2. Active flow determination, path set generation, and overlapping path elimination are performed according to patented procedures;

[0133] 3. Collect and process the total number of links in the resulting path.

[0134] Experiment 2: Comparison of Path Planning Algorithms

[0135] Purpose of the experiment:

[0136] Comparison of repeated links between the in-band telemetry path planning algorithm based on segment routing and the path generated by the traditional Euler path algorithm

[0137] Experimental steps:

[0138] 1. Set up an INT simulation environment in Mininet, use Python to randomly generate topological structure graphs with different numbers of vertices, and determine the correspondence between the starting point and end point of each flow;

[0139] 2. Active flow determination, path set generation, and overlapping path elimination are performed according to patented procedures;

[0140] 3. Arrange the same network topology in Mininet;

[0141] 4. Generate a set of planned paths using the traditional Euler path algorithm;

[0142] 5. Compare the number of repeated links in the path sets generated by the two algorithms.

[0143] It should be noted that the above embodiments are merely preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Equivalent replacements or substitutions made on the basis of the above technical solutions all fall within the scope of protection of the present invention.

Claims

1. An in-band telemetry path planning algorithm based on segment routing, It is characterized in that The algorithm consists of the following steps: (1) Modeling network topology and network flow. Modeling network topology as an undirected connected graph G, and modeling flow as a six-tuple; (2) Based on the existing and historical information of the flow in the network, determine whether the flow is active. Convert all original paths that are determined to be active flows into connected subgraphs G ′ ; (3) Determine the connected graph G ′ Under the condition that the starting point and end point of each flow correspond to each other, the optimized Euler path algorithm is used to find the optimal path in the connected graph G. ′ Generate the path result set E′ and plan a network telemetry path plan with less overlap; (4) Using the overlapping path elimination algorithm, the overlapping paths are migrated by adding auxiliary points, introducing new nodes and new links, thereby reducing redundancy and generating the final path planning result.

2. The in-band telemetry path planning algorithm based on segment routing according to claim 1, It is characterized in that The specific method of step (1) is as follows: (2.1) Model the network topology as an undirected connected graph G = (V, E); (2.2) The flow in the network is abstracted into a six-tuple f = (tuple, r, flag, t s ,t e ,path); (2.3) In the segment routing environment, path can be expressed as path(f) = (SID[n],…,SID[2],SID[i]) = (v 1 ,v 2 ,…,v n ), where v 1 ∈V,(v 1 ,v 1 )∈E; ( 2.4) Time t e -t s All active flows constitute the set F = {f 1 ,f 2 ,…,f n }.

3. The in-band telemetry path planning algorithm based on segment routing according to claim 1, It is characterized in that The design method for active flow judgment and connected subgraph generation in step (2) is as follows: (3.1) Set the flow activity threshold; (3.2) Based on historical INT data, determine whether the activity of the flow exceeds the threshold; (3.3) At the cost of generating repeated paths, transform the original paths of all active flows into a connected subgraph G′ of graph G; (3.4) Determine the correspondence between the starting point and the end point of each flow in the connected subgraph G′.

4. The in-band telemetry path planning algorithm based on segment routing according to claim 1, It is characterized in that The specific method of the path planning algorithm in step (3) is as follows: (4.1) Find the path with the largest coverage rate between the odd points in the graph G′; (4.2) Remove the planned path; (4.3) Find a path from the starting point that is as non-repetitive as possible; (4.4) Generate the path planning result set E′.

5. The in-band telemetry path planning algorithm based on segment routing according to claim 1, It is characterized in that The specific method of the overlapping path elimination algorithm in step (4) is as follows: (5.1) Count all overlapping paths; (5.2) Select auxiliary nodes for overlapping paths; (5.3) Use auxiliary nodes to construct new paths; (5.4) Generate the path result set R after removing redundancy.

Citation Information

Patent Citations

  • Network telemetering method based on SRv6

    CN114050994A

  • Traffic adjusting and optimizing method and system based on IPv6 segment routing and in-band network telemetering

    CN115733783A

  • Network measurement method and device based on SRv6 and INT

    CN115766552A

  • Software defined networking with en-route computing

    US20220021608A1

Cited By

  • SRV6 path confusion method based on pseudo-random curvature diffusion

    CN119484388A