Multilink test method and system based on SD-WAN

Abstract

The invention provides an SD-WAN-based multi-link test method and system, and relates to the technical field of computer networks, and the method comprises the steps: transmitting a detection message to a second to-be-tested device through a first to-be-tested device, and receiving a feedback message returned by the second to-be-tested device for the detection message; determining respective link quality data of the plurality of links to be detected according to the detection message and the feedback message; based on a clustering algorithm model obtained by pre-training, performing clustering processing on the respective link quality data of the plurality of links to be tested to obtain a plurality of clusters for representing the link quality and respective convergence areas of the plurality of clusters; and determining the link quality data which do not belong to the convergence area as abnormal data, and performing anomaly analysis on the abnormal data. According to the invention, a bidirectional communication mechanism of the detection message and the feedback message is adopted, the complexity of an SD-WAN test environment is effectively simplified, and the test cost is reduced.

Description

Technical Field

[0001] The present disclosure relates to the field of computer network technology, and in particular to a multi-link testing method and system based on SD-WAN. Background Art

[0002] In the digital transformation process of modern enterprises, SD-WAN (Software Defined Wide Area Network), as an important network architecture, has been widely used in scenarios such as branch office interconnection, data center connection, cloud service access, and cross-border office. In a typical SD-WAN deployment, the corporate headquarters or central node needs to establish connections with multiple branches, cloud service providers, and data centers, which usually involves dozens or even hundreds of link detections.

[0003] However, each node often needs to be equipped with a UCPE (Unified Edge Computing Platform) box, which makes the test environment complex and costly. As the business of enterprises expands, the expansion of link requirements has become the norm, which makes enterprises face huge challenges in network management and maintenance. The deployment and management of each UCPE box not only increases the hardware cost, but also makes the network architecture more complex. Especially when real-time monitoring and performance evaluation of each link are required, enterprises need to invest more resources and manpower for maintenance, which makes the test environment complex and costly. Summary of the invention

[0004] The embodiments of the present disclosure provide a multi-link testing method and system based on SD-WAN, aiming to solve the problems existing in the above-mentioned background technology.

[0005] In order to solve the above technical problems, the present disclosure is implemented as follows: In a first aspect, an embodiment of the present disclosure provides a multi-link testing method based on SD-WAN, the method comprising: Sending a detection message to a second device under test through a first device under test, and receiving a feedback message returned by the second device under test in response to the detection message, wherein the first device under test and the second device under test are respectively configured with a plurality of links under test; Determining link quality data of each of the multiple links to be tested according to the detection message and the feedback message; Based on the clustering algorithm model obtained by pre-training, clustering processing is performed on the link quality data of each of the multiple links to be tested to obtain multiple clusters for characterizing the link quality and convergence regions of each of the multiple clusters; Link quality data that does not belong to the convergence area is determined as abnormal data, and an abnormality analysis is performed on the abnormal data.

[0006] Optionally, after receiving a feedback message returned by the second device under test in response to the detection message, the method further includes: Determining, according to the detection message and the feedback message, an actual bandwidth of each of the multiple links to be tested; For any link to be tested, comparing the actual bandwidth of the link to be tested with a preset first threshold; When the actual bandwidth is greater than or equal to a preset first threshold, sending a data stream to the first device under test for the link under test through a NETPERF stream-breaking tool of the second device under test; The current remaining bandwidth of the link to be tested is determined by using the SAR traffic monitoring tool of the first device to be tested, and the current remaining bandwidth is used as the link quality data of the link to be tested.

[0007] Optionally, the method further comprises: Determine the CPU occupancy rate and memory usage rate of each of the multiple links to be tested through the SAR traffic monitoring tool of the first device to be tested; From the multiple links to be tested, determine a CPU-exclusive link to be tested whose CPU occupancy rate is higher than a preset second threshold, and determine a memory leak link to be tested whose memory usage rate is higher than a preset third threshold; For a link under test that is exclusively used by the CPU or a link under test that has a memory leak, the scheduling priority and resource allocation strategy of the link under test are correspondingly adjusted in the QOS yang files of the first device under test and the second device under test.

[0008] Optionally, the method further comprises: In the configuration files of the first device under test and the second device under test, respectively, the link uuid number is increased to a preset link upper limit, wherein the link uuid number corresponds one-to-one to the multiple links under test; Monitoring the link quality data of each of the plurality of links under test by using the SAR traffic monitoring tool of the first device under test, wherein the link quality data includes: delay, jitter and packet loss rate; Determining the interface status between the first device under test and the second device under test according to the link quality data of each of the multiple links under test; When the interface status indicates that the interface is blocked, the scheduling priorities and resource allocation strategies of the multiple links to be tested are correspondingly adjusted in the QOS yang files of the first device to be tested and the second device to be tested.

[0009] Optionally, the clustering algorithm model is pre-trained according to the following steps: Collecting a plurality of historical link quality data of a plurality of links to be tested in different time periods, and inputting the plurality of historical link quality data as training samples into the clustering algorithm model; Randomly selecting N historical link quality data from the plurality of historical link quality data as the first clustering centroid of each of the N clusters, where N is an integer greater than 1; Calculating the distance between each historical link quality data and the centroid of a cluster, and assigning the plurality of historical link quality data to the nearest cluster; Iteratively update the cluster center of gravity according to the average value of the historical link quality data in the same cluster until the change difference between two consecutive updated cluster centers of gravity is less than a preset change threshold, and obtain N converged clusters; According to the cluster centroids of the N converged clusters, the convergence regions of the N clusters are determined.

[0010] Optionally, after clustering the link quality data of each of the plurality of links to be tested based on the clustering algorithm model obtained through pre-training, the method further includes: Determine the cluster center of gravity of each of the multiple clusters, and a link quality indicator corresponding to each cluster center of gravity, wherein the link quality indicator is a one-dimensional indicator; Mapping the link quality data of each of the multiple links to be tested to the cluster center of gravity of the cluster to which they belong, to obtain the link quality indicators of each of the multiple links to be tested, wherein the link quality data at least includes: delay, jitter and packet loss rate; The method further comprises: Monitoring whether there is a faulty link among the multiple links to be tested; In the case where there is a faulty link, determining the cluster center of gravity of the cluster to which the faulty link belongs; Determine, from the cluster to which the faulty link belongs, a backup link that is closest to the link quality indicator of the abnormal link and whose current remaining bandwidth meets the preset bandwidth requirement; The failed link is switched to the standby link.

[0011] Optionally, after clustering the link quality data of each of the plurality of links to be tested based on the clustering algorithm model obtained through pre-training, the method further includes: Collect multiple historical link quality data of multiple links to be tested in different time periods to form a historical data set; Determining, according to the historical data set, a link quality evolution trend of each of the multiple links to be tested; Based on the clustering results of the link quality data and the link quality evolution trend, potential faulty links whose link quality data exceeds a set tolerance range for multiple consecutive time periods are identified, and result-oriented analysis is performed on the potential faulty links.

[0012] In a second aspect, an embodiment of the present disclosure provides a multi-link test system based on SD-WAN, which is applied to the steps of the multi-link test method based on SD-WAN as described above, and the system includes: a first device under test, a second device under test, and a PC device, wherein the first device under test and the second device under test are respectively configured with a plurality of links under test; the first device under test is connected to the PC device via a LAN port, and the first device under test and the second device under test are connected via a WAN port; The first device under test is used to send a detection message to the second device under test, receive a feedback message returned by the second device under test in response to the detection message, and report link quality data of each of the multiple links under test to the PC device according to the detection message and the feedback message; The second device under test is used to receive the detection message and send the feedback message to the first device under test; The PC device performs clustering processing on the link quality data of each of the multiple links to be tested based on the clustering algorithm model obtained through pre-training, and the result of the clustering processing represents the link quality characteristics of each of the multiple links to be tested.

[0013] Optionally, the first device under test accesses and obtains the IP address of the first device under test through the PC device; the WAN port of the second device under test is set with the same network segment as the first device under test but with a different IP address; The first device under test is added with a first routing detail, and the first routing detail accesses the IP address of the second device under test through the WAN port of the second device under test; The second device under test is added with a second routing detail, and the second routing detail accesses the IP address of the first device under test through the WAN port of the second device under test; The system also includes a WAN port verification module for verifying whether the WAN port of the first device under test can ping the IP address of the WAN port of the second device under test, and verifying whether the WAN port of the second device under test can ping the IP address of the WAN port of the first device under test.

[0014] Optionally, the first device under test is configured with a first configuration file, and the second device under test is configured with a second configuration file; The first configuration file is used to create the same number of link uuid numbers for the multiple links to be tested, and to set an end detection IP that is the same as the IP address of the second device to be tested, wherein the link uuid number corresponds one-to-one to the multiple links to be tested; The second configuration file is used to create a link uuid number that corresponds one-to-one to the link uuid number configured in the first configuration file, and to set an end detection IP that is the same as the IP address of the first device under test.

[0015] The technical solution provided by the embodiments of the present disclosure brings at least the following beneficial effects: The present disclosure adopts a two-way communication mechanism of detection messages and feedback messages, effectively simplifies the complexity of the SD-WAN test environment, reduces the test cost, and accurately obtains the quality data of each link under test through real-time data exchange between the first device under test and the second device under test, thereby using the pre-trained clustering algorithm model to cluster the link quality data, which can quickly identify the normal and abnormal states of the link, reducing manual intervention and complex manual analysis processes. It can be seen that the present disclosure significantly improves the efficiency of SD-WAN testing, reduces the investment of enterprises in network management and maintenance, and provides strong support for the digital transformation of enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 This is a schematic diagram of a typical application scenario of SD-WAN provided by an embodiment of the present disclosure; Figure 2 It is a structural block diagram of a multi-link test system based on SD-WAN provided by an embodiment of the present disclosure; Figure 3 is a schematic diagram of a network environment configuration process of a first device under test and a second device under test in one embodiment of the present disclosure; Figure 4 It is a schematic diagram of the steps of a multi-link testing method based on SD-WAN provided by an embodiment of the present disclosure; Figure 5 is a schematic diagram of the interaction steps between the first device under test and the second device under test in one embodiment of the present disclosure; Figure 6 is a flow chart of clustering model training and application in one embodiment of the present disclosure; Figure 7 is a data distribution diagram of a clustering result in one embodiment of the present disclosure; Figure 8 It is a data distribution diagram of another clustering result in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0018] The following will be combined with the drawings in the embodiments of the present disclosure to clearly and completely describe the technical solutions in the embodiments of the present disclosure. Obviously, the described embodiments are part of the embodiments of the present disclosure, rather than all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present disclosure.

[0019] Figure 1 This is a schematic diagram of a typical SD-WAN application scenario provided by an embodiment of the present disclosure. Figure 1 In a typical application scenario of SD-WAN, the central node, as the core of the network, is responsible for efficient communication and data transmission with multiple branch nodes. These branch nodes can include cloud service providers, data centers, and branch offices in various geographically dispersed locations. Through SD-WAN technology, enterprises can achieve flexible network connections, optimize data flow, and improve network reliability and performance. In this network architecture, the central node communicates with multiple branch nodes in real time through networking connections to ensure fast data transmission between different locations. For example, a branch node can be a cloud service platform that provides the computing and storage resources required by the enterprise; it can also be a data center that is responsible for storing and processing large amounts of data; in addition, the branch office is the front line of the company's daily operations. Through the SD-WAN connection, employees can access the central database and applications.

[0020] In related technologies, each node often needs to be equipped with a UCPE (unified edge computing platform) box to support the functions of SD-WAN. However, each branch node needs to deploy a UCPE box separately, which not only increases the initial investment cost, but also leads to an increase in subsequent maintenance and upgrade costs. As the number of branch nodes increases, the number of UCPE devices that enterprises need to manage and maintain will also increase, making the network architecture more complicated. Each UCPE box needs to be configured, monitored, and troubleshooted, which increases the complexity of network management.

[0021] In view of the defects of the above-mentioned related technologies, the present disclosure aims to reduce the hardware requirements of each node. Through centralized management and intelligent scheduling, effective management and monitoring of branch links or nodes can be achieved without the need to deploy physical devices separately. Figure 2 This is a structural block diagram of a multi-link test system based on SD-WAN provided by an embodiment of the present disclosure, and the system is applied to the test method described below. The system includes: a first device under test, a second device under test, and a PC device, and the first device under test and the second device under test are respectively configured with multiple links under test.

[0022] The first device under test acts as the main node of the network and is responsible for actively initiating detection requests. The second device under test acts as a branch node in the network and is mainly responsible for receiving detection messages from the first device under test and returning feedback messages. The link under test refers to the network connection configured between the first device under test and the second device under test, which can refer to a physical connection or a virtual network connection (such as optical fiber, Ethernet, VPN, etc.), which is respectively configured on the first device under test and the second device under test. Each link under test represents an independent network path, and performance tests are performed through these links to evaluate their link quality.

[0023] The first device under test is connected to the PC device via a LAN port, and the first device under test and the second device under test are connected via a WAN port.

[0024] Figure 3 FIG. 1 is a schematic diagram of a network environment configuration process of a first device under test and a second device under test in an embodiment of the present disclosure. Figure 3 As shown, the first device under test is connected to the PC device through its LAN (local area network) port. Through the LAN port, the first device under test can transmit data with the PC device. The PC device can access the management interface of the first device under test through the network to perform operations such as setting, monitoring traffic, and viewing status. During the test, the PC device is used as a console to send test instructions, collect data, and analyze results.

[0025] The first device under test and the second device under test are connected via a WAN (Wide Area Network) port. The WAN port connection is used to interconnect the first device under test with the second device under test to form a wide area network environment, simulating a real network environment to test the performance of the device under different network conditions. Through the WAN port, the first device under test and the second device under test can send and receive data packets to each other.

[0026] It should be noted that if the user needs to control the second device under test through the PC device to implement message exchange and data reporting, the second device under test can also be connected to the PC device through a USB port.

[0027] The first device under test is used to send a detection message to the second device under test, receive a feedback message returned by the second device under test in response to the detection message, and report link quality data of each of the multiple links under test to the PC device according to the detection message and the feedback message.

[0028] In order to evaluate and monitor the quality of the link, the first device under test sends a detection message to the second device under test. The detection message includes the timestamp and message count of the sender in the service cycle. After receiving the detection message, the second device under test performs corresponding processing and returns a feedback message to the first device under test. The first device under test can obtain link quality data representing the link performance quality based on the detection message and the feedback message.

[0029] The second device under test is used to receive the detection message, and send the feedback message to the first device under test.

[0030] The feedback message includes the timestamp and message count of the receiving end within the service cycle.

[0031] The following is a specific example to illustrate how to report the link quality data of the link to be tested to the PC device based on the detection message and the feedback message. Assume that the first device to be tested Device A sends a detection message to the second device to be tested Device B, and the content includes: timestamp (T1): 2023-10-01 10:00:00; message count (Seq): 1 (indicating that this is the first detection message). Device A sends a detection message to Device B. After Device B receives the detection message, it records the receiving timestamp (T2) and processes it. Device B returns a feedback message to Device A, and the content includes: receiving timestamp (T2): 2023-10-01 10:00:01; message count (Seq): 1 (indicating that this is the feedback corresponding to the detection message). After the first device to be tested receives the feedback message, the link quality data can be obtained according to the message content of the detection message and the feedback message. In the embodiment of the present disclosure, the link quality data includes: delay, jitter and packet loss rate. Assume that within a service cycle, Device A sends 10 probe messages, and Device B only successfully receives 8 feedback messages. Packet loss rate = (total number of sent messages - total number of received feedback messages) / total number of sent messages × 100%. Total number of sent messages = 10; total number of received feedback messages = 8; therefore, packet loss rate = (10 - 8) / 10 × 100% = 20%. Assume that Device A sends 3 probe messages within a service cycle, and records the following timestamps: T1 = 2023-10-01 10:00:00 (feedback time T2 = 2023-10-01 10:00:01); T3 = 2023-10-01 10:00:02 (feedback time T4 = 2023-10-01 10:00:03); T5 = 2023-10-01 10:00:04 (feedback time T6 = 2023-10-01 10:00:05); calculate the delay of each detection message: First message delay = 1 second; second message delay = 1 second; third message delay = 1 second Jitter is the variation of delay. The standard deviation of delay is calculated as: Mean delay = (1 + 1 + 1) / 3 = 1 second; Standard Deviation = = 0 (because all delays are the same), where N is the number of packets = 3. Therefore, the jitter in this example is 0 seconds.

[0032] Through the above steps, the link quality data obtained are: delay 1 second, jitter 0 seconds, and packet loss rate 20%.

[0033] In this embodiment, in order to ensure normal testing and data transmission of the link, the first device under test and the second device under test need to consider the influence of the network address translation (NAT) setting when configuring. If the first device under test has a NAT configuration, it may cause the data packet to be modified when passing through the device, thereby affecting the performance test results of the link. Therefore, in this case, it is necessary to delete the NAT configuration of the first device under test to ensure that the data can be transmitted directly through the device and maintain the original state of the link. Similarly, if the second device under test also has a NAT configuration, it also needs to be deleted.

[0034] The PC device performs clustering processing on the link quality data of each of the multiple links to be tested based on the clustering algorithm model obtained through pre-training, and the result of the clustering processing represents the link quality characteristics of each of the multiple links to be tested.

[0035] By collecting link quality data from the first device under test and the second device under test, the PC device uses a clustering algorithm to analyze and classify the link quality data. The result of the clustering process can effectively characterize the quality characteristics of each link under test, so that network managers can intuitively identify the performance differences of different links, discover potential abnormal links, and make corresponding optimizations and adjustments based on this information. In the test method below, the clustering algorithm and its training process will be described in detail.

[0036] The present disclosure can comprehensively evaluate the quality characteristics of the network link, including link quality data such as latency, jitter, and packet loss rate, through the collaborative work of the first device under test and the second device under test, as well as the cluster analysis of the PC device. It can be seen that the present disclosure can achieve real-time link quality monitoring using only three devices, and can also identify performance differences and potential anomalies of the link through data clustering, which helps to optimize the network configuration, improve the stability and transmission efficiency of the network, and thus meet the high performance requirements in complex network environments.

[0037] In an optional implementation, the first device under test accesses and obtains the IP address of the first device under test through the PC device; the WAN port of the second device under test is set with an IP address that is in the same network segment as the first device under test but different from that of the first device under test; the first device under test is added with a first routing detail, and the first routing detail accesses the IP address of the second device under test through the WAN port of the second device under test; the second device under test is added with a second routing detail, and the second routing detail accesses the IP address of the first device under test through the WAN port of the second device under test.

[0038] like Figure 3 As shown, in this embodiment, the first device under test is connected to the PC device, and obtains its own IP address through the LAN port using a network protocol (such as DHCP or static configuration). The PC device, as a management terminal, accesses the network settings of the first device under test through a command line or a graphical interface to obtain its assigned IP address. The WAN port of the second device under test is configured to be in the same network segment as the first device under test, but with a different IP address. In other words, the IP addresses of the two devices are in the same subnet and can recognize and communicate with each other. For example, if the IP address of the first device under test is 192.168.1.2, the WAN port of the second device under test can be set to 192.168.1.3. The first device under test adds a routing detail in its routing table, pointing to the IP address of the second device under test. In other words, the first device under test can send a data packet to the WAN port of the second device under test through its WAN port. The first routing detail ensures the correctness of the data flow direction, so that the first device under test can effectively access the resources of the second device under test or perform link quality testing. Similarly, the second device under test adds a second routing detail in its routing table, pointing to the IP address of the first device under test, to ensure that the second device under test can access the resources of the first device under test through its WAN port.

[0039] The system also includes a WAN port verification module for verifying whether the WAN port of the first device under test can ping the IP address of the WAN port of the second device under test, and verifying whether the WAN port of the second device under test can ping the IP address of the WAN port of the first device under test.

[0040] The WAN port verification module verifies the reachability between the WAN ports of the two devices under test by sending a PING request. PING is a network diagnostic tool that tests the validity of the network connection by sending an ICMP (Internet Control Message Protocol) echo request to the target IP address and waiting for an echo response. First, the WAN port verification module sends a PING request from the WAN port of the first device under test to the IP address of the WAN port of the second device under test. Specifically, the WAN port IP address of the second device under test is determined, a PING request is sent to the IP address through the ICMP protocol, and an echo response from the second device under test is waited for. If the WAN port of the second device under test can successfully receive the PING request and return a response, the WAN port verification module will record the connection as "reachable". If no response is received, it is recorded as "unreachable". Then, the WAN port verification module will perform the opposite operation and send a PING request from the WAN port of the second device under test to the IP address of the WAN port of the first device under test. Specifically, the WAN port IP address of the first device under test is determined, a PING request is sent to the IP address through the ICMP protocol, and an echo response from the first device under test is waited for. Similarly, if the WAN port of the first device under test can successfully receive the PING request and return a response, the WAN port verification module will record the connection as "reachable". If no response is received, it will be recorded as "unreachable".

[0041] In an optional embodiment, the first device under test is configured with a first configuration file, and the second device under test is configured with a second configuration file; the first configuration file is used to create the same number of link uuid numbers for the multiple links under test, and set the end detection IP that is the same as the IP address of the second device under test, wherein the link uuid number corresponds one-to-one to the multiple links under test; the second configuration file is used to create link uuid numbers that correspond one-to-one to the link uuid numbers configured by the first configuration file, and set the end detection IP that is the same as the IP address of the first device under test.

[0042] like Figure 3As shown, in the first configuration file or the second configuration file, a unique link UUID number is generated for each link to be tested. UUID (Universally Unique Identifier) ​​is a standard identifier that ensures that each link is unique in the system, and the link UUID number corresponds one-to-one to the link to be tested. In the first link configuration file, an end detection IP that is the same as the IP address of the second device to be tested is set. In other words, the first device to be tested will use the IP address of the second device to be tested as the detection target in order to perform link quality testing. Similarly, in the second link configuration file, an end detection IP that is the same as the IP address of the first device to be tested is set. The second device to be tested will use the IP address of the first device to be tested as the detection target.

[0043] In an optional implementation, the QOS yang file of the first device under test is configured with attribute information of each of the multiple links under test for the link uuid number, and the first device under test monitors the configuration of the QOS yang file through a monitoring switch, and the attribute information includes link priority and bandwidth limitation; the QOS yang file of the second device under test is configured with attribute information of each of the multiple links under test for the link uuid number, and the second device under test monitors the configuration of the QOS yang file through a monitoring switch.

[0044] like Figure 3 As shown, in this embodiment, the QOS (Quality of Service) yang files of the first device under test and the second device under test are configured for their respective link UUID numbers to define the attribute information of multiple links under test. The attribute information includes the priority and bandwidth limit of the link to ensure the reasonable allocation and optimized use of network resources. The first device under test monitors the configuration status of its QOS yang file in real time through the monitoring switch so as to adjust the quality of service settings of the link in time; similarly, the second device under test also monitors the corresponding QOS yang file configuration through its monitoring switch.

[0045] The test system provided in this embodiment can be applied to execute the steps of the method described below after being configured in the above network environment. Figure 4 FIG. 1 is a schematic diagram of a multi-link testing method based on SD-WAN provided by an embodiment of the present disclosure. Figure 4 As shown, the method includes: Step S101: Send a detection message to a second device under test through a first device under test, and receive a feedback message returned by the second device under test in response to the detection message. The first device under test and the second device under test are respectively configured with a plurality of links under test.

[0046] The first device under test sends a detection message to the second device under test. The detection message includes the timestamp and message count of the sender in the service cycle. After receiving the detection message, the second device under test performs corresponding processing and returns a feedback message to the first device under test.

[0047] Step S102: determining link quality data of each of the plurality of links to be tested according to the detection message and the feedback message.

[0048] As mentioned above, the link quality data is an indicator used to evaluate the performance of the link to be tested. In this embodiment, the link quality data at least includes delay, jitter, packet loss rate and current remaining bandwidth.

[0049] Figure 5 is a schematic diagram of the interaction steps between the first device under test and the second device under test in one embodiment of the present disclosure, such as Figure 5 As shown, step S101 to step S102 are executed cyclically when facing the multi-link scenario in this embodiment until a preset clock cycle condition is met.

[0050] Step S103 , based on the clustering algorithm model obtained through pre-training, clustering processing is performed on the link quality data of each of the plurality of links to be tested to obtain a plurality of clusters for characterizing link quality and convergence regions of the plurality of clusters.

[0051] The clustering algorithm is an unsupervised learning method that aims to group objects in a data set according to their characteristics. The pre-trained clustering algorithm model has learned how to identify and group similar link quality characteristics through historical link quality data.

[0052] The collected link quality data is input into the clustering algorithm model, which analyzes and groups the link quality data. Similar link quality features are classified into the same category, while different features are classified into different clusters. Through clustering processing, multiple clusters are generated, each cluster represents a group of links with similar link quality features.

[0053] Each cluster has its own convergence region, which indicates the normal range of link quality data within the cluster. If a data point falls within the convergence region, it indicates that its quality is normal; if it falls outside, it indicates that the link is abnormal.

[0054] Step S104: determining link quality data that does not belong to the convergence area as abnormal data, and performing abnormality analysis on the abnormal data.

[0055] Check the quality data of all links to be tested to determine which data points are not within the convergence region of their corresponding clusters. These data points that are not within the convergence region are marked as abnormal data, indicating that the quality performance of these links is significantly different from normal conditions. For the identified abnormal data, perform detailed abnormality analysis. Specifically, you can analyze the causes of the abnormal data and find out the specific factors that cause the link quality to deteriorate by combining information such as network topology, device status, and traffic monitoring. You can also observe the time series changes of abnormal data to determine whether it is an isolated event or a persistent problem, and analyze its frequency and pattern of occurrence. You can also compare abnormal links with normal links to find differences and help identify potential root causes of failures.

[0056] The present disclosure adopts a two-way communication mechanism of detection messages and feedback messages, effectively simplifies the complexity of the SD-WAN test environment, reduces the test cost, and accurately obtains the quality data of each link under test through real-time data exchange between the first device under test and the second device under test, thereby using the pre-trained clustering algorithm model to cluster the link quality data, which can quickly identify the normal and abnormal states of the link, reducing manual intervention and complex manual analysis processes. It can be seen that the present disclosure significantly improves the efficiency of SD-WAN testing, reduces the investment of enterprises in network management and maintenance, and provides strong support for the digital transformation of enterprises.

[0057] In an optional implementation, after receiving a feedback message returned by the second device under test in response to the detection message, the method further includes: Determining, according to the detection message and the feedback message, an actual bandwidth of each of the multiple links to be tested; For any link to be tested, comparing the actual bandwidth of the link to be tested with a preset first threshold; When the actual bandwidth is greater than or equal to a preset first threshold, sending a data stream to the first device under test for the link under test through a NETPERF stream-breaking tool of the second device under test; The current remaining bandwidth of the link to be tested is determined by using the SAR traffic monitoring tool of the first device to be tested, and the current remaining bandwidth is used as the link quality data of the link to be tested.

[0058] Here, the probe message can include the packet size in addition to the timestamp and message count mentioned above. Figure 5, for the link with bandwidth requirements, the NETPERF streaming tool of the second device under test is used to send data streams to the first device under test for the link under test. For example, assuming that the first device under test Device A and the second device under test Device B have been successfully configured, a probe message is sent to Device B for a certain link under test. For example, Device A sends a data packet of 1000 bytes with a timestamp of T1. After receiving the probe message, Device B processes the message and returns a feedback message. The feedback message contains information such as the timestamp and receiving status of the received data packet. Assume that Device B successfully receives the data packet at the T2 timestamp (for example, T2 = T1 + 0.1 seconds) and returns a feedback message to Device A. Device A calculates the actual bandwidth based on the probe message and the feedback message. Assuming that the information contained in the feedback message indicates that the data packet is successfully transmitted within 0.1 seconds, the actual bandwidth can be obtained as:

[0059] Assume that the preset first threshold is 5 kbps. Device A compares the actual bandwidth 10 kbps with the preset threshold 5 kbps. Because 10 kbps is greater than or equal to 5 kbps, it means that the link to be tested has bandwidth requirements. Device B uses the NETPERF traffic tool to send a series of data streams to Device A for the link. For example, Device B sends data streams to Device A at a rate of 5 kbps for a period of time. At the same time, Device A uses the SAR traffic monitoring tool (SystemActivity Reporter) to monitor the traffic of the link. Assume that Device A detects that the current total bandwidth of the link to be tested is 20 kbps, and the bandwidth in use is 5 kbps (data stream from Device B). The current remaining bandwidth is: current remaining bandwidth = total bandwidth − bandwidth in use = 20kbps − 5kbps = 15kbps. Device A records the current remaining bandwidth of 15kbps as the link quality data of the link to be tested, which will be used for subsequent performance evaluation and monitoring.

[0060] Through the above steps, the first device under test and the second device under test can effectively evaluate and monitor the link performance between them, helping the network administrator to understand the bandwidth usage of the link in real time according to the bandwidth demand of the link, so as to take timely measures to ensure the efficient operation of the network.

[0061] In an optional embodiment, the method further includes: Determine the CPU occupancy rate and memory usage rate of each of the multiple links to be tested through the SAR traffic monitoring tool of the first device to be tested; From the multiple links to be tested, determine a CPU-exclusive link to be tested whose CPU occupancy rate is higher than a preset second threshold, and determine a memory leak link to be tested whose memory usage rate is higher than a preset third threshold; For a link under test that is exclusively used by the CPU or a link under test that has a memory leak, the scheduling priority and resource allocation strategy of the link under test are correspondingly adjusted in the QOS yang files of the first device under test and the second device under test.

[0062] The SAR traffic monitoring tool of the first device under test is used to obtain the CPU occupancy rate and memory usage rate of each link under test in real time. The SAR traffic monitoring tool regularly monitors the resource consumption of the first device under test and provides the real-time consumption of the CPU and memory corresponding to each link under test. For example: Link A under test: CPU usage is 85%, and memory usage is 60%.

[0063] Link B under test: CPU usage is 95%, and memory usage is 50%.

[0064] Link C under test: CPU usage is 40%, and memory usage is 85%.

[0065] According to the preset second threshold (for example, 90%), abnormal links are identified, and links with CPU occupancy higher than the preset second threshold are determined as CPU-exclusive links to be tested. Links with memory usage higher than the preset third threshold (for example, 80%) are determined as memory leak links to be tested. For link B to be tested, the CPU occupancy is 95%, exceeding the threshold of 90%, and is marked as a CPU-exclusive link to be tested. For link C to be tested, the memory usage is 85%, exceeding the threshold of 80%, and is marked as a memory leak link to be tested.

[0066] The memory leak link and CPU-exclusive link are the links that cause system performance bottlenecks. They are adjusted in the QOS yang files of the first device under test and the second device under test respectively. For the CPU-exclusive link under test, reduce the scheduling priority of the link and reduce resource allocation. Prevent CPU overload from affecting the normal operation of other links. For the memory leak link, increase the scheduling priority of the link to ensure that sufficient resources are allocated to cope with memory pressure. If the problem persists, you can set a trigger alarm or restart mechanism.

[0067] In this example, for link B to be tested (CPU exclusive), modify the QOS yang files of the devices on both sides and reduce the scheduling priority from high to low. Limit the upper limit of the CPU resource usage of this link to 80%. For link C to be tested (memory leak), increase its scheduling priority to the highest, dynamically adjust the memory allocation strategy, reserve more memory space, and prevent task interruption.

[0068] The present invention automatically adjusts the link scheduling strategy by real-time monitoring of resource usage, avoiding single-point exhaustion of CPU and memory resources, ensuring the stability and performance of the overall system, and improving the operating efficiency of key links and reducing link delay and packet loss rate through link priority adjustment.

[0069] In an optional embodiment, the method further includes: In the configuration files of the first device under test and the second device under test, respectively, the link uuid number is increased to a preset link upper limit, wherein the link uuid number corresponds one-to-one to the multiple links under test; Monitoring the link quality data of each of the plurality of links under test by using the SAR traffic monitoring tool of the first device under test, wherein the link quality data includes: delay, jitter and packet loss rate; Determining the interface status between the first device under test and the second device under test according to the link quality data of each of the multiple links under test; When the interface status indicates that the interface is blocked, the scheduling priorities and resource allocation strategies of the multiple links to be tested are correspondingly adjusted in the QOS yang files of the first device to be tested and the second device to be tested.

[0070] In the configuration files of the first device under test and the second device under test, the UUID (globally unique identifier) ​​of the link is increased to the preset link upper limit respectively. Each UUID corresponds to a link under test one by one to ensure that the system can accurately identify and manage each link. For example, assuming that the preset link upper limit is 10, a unique UUID will be assigned to each link, such as uuid1, uuid2, ..., uuid10. Use the SAR traffic monitoring tool of the first device under test to monitor the link quality data of each link under test. The calculation and monitoring process of latency, jitter and packet loss rate has been described in detail in the previous article and will not be repeated here.

[0071] According to the monitored link quality data, the interface status between the first device under test and the second device under test is determined. The interface status includes normal, congested, and faulty. If the monitored link quality data (such as high latency, high jitter, or high packet loss rate) indicates that the interface is blocked, the interface status will be identified as blocked. Assume that the monitoring results are as follows: Link uuid1: latency = 20ms, jitter = 5ms, packet loss rate = 0%; Link uuid2: latency = 50ms, jitter = 10ms, packet loss rate = 2%; Link uuid3: delay = 30ms, jitter = 3ms, packet loss rate = 1%; it can be seen that the packet loss rate of link uuid2 is 2%, and the delay and jitter are also high, and the interface status is judged to be "congested". When the interface status indicates that the interface is congested, the scheduling priority and resource allocation strategy of multiple links under test will be adjusted in the QOS yang files of the first device under test and the second device under test respectively. Increase the scheduling priority of certain key links to ensure the smooth transmission of important data streams. Allocate more bandwidth resources to high-priority links and reduce the bandwidth allocation of other low-priority links. For example, the priority of link uuid1 may be raised from the normal level to the high priority level, while reducing the bandwidth allocation of link uuid2 to alleviate congestion.

[0072] Through the above steps, the quality of multiple links to be tested can be effectively monitored and managed, the congestion status of the interface can be identified in time, and the link priority and resource allocation strategy can be adjusted according to the actual situation. This dynamic adjustment mechanism can improve the overall performance and stability of the network and ensure the smooth progress of key services.

[0073] In an optional implementation, the clustering algorithm model is pre-trained according to the following steps: Collecting a plurality of historical link quality data of a plurality of links to be tested in different time periods, and inputting the plurality of historical link quality data as training samples into the clustering algorithm model; Randomly selecting N historical link quality data from the plurality of historical link quality data as the first clustering centroid of each of the N clusters, where N is an integer greater than 1; Calculating the distance between each historical link quality data and the centroid of a cluster, and assigning the plurality of historical link quality data to the nearest cluster; Iteratively update the cluster center of gravity according to the average value of the historical link quality data in the same cluster until the change difference between two consecutive updated cluster centers of gravity is less than a preset change threshold, and obtain N converged clusters; According to the cluster centroids of the N converged clusters, the convergence regions of the N clusters are determined.

[0074] Figure 6 This is a flowchart of clustering model training and application in one embodiment of the present disclosure, please refer to Figure 6In this embodiment, the clustering algorithm model is an unsupervised learning tuning type K-means algorithm model. During its pre-training process, historical link quality data of multiple links to be tested in different time periods are collected. The historical link quality data may include: delay, jitter and packet loss rate. For example, it is assumed that the following historical link quality data (each data contains delay, jitter and packet loss rate) are collected in different time periods: Data 1: latency = 20ms, jitter = 2ms, packet loss rate = 0%; Data 2: latency = 30ms, jitter = 5ms, packet loss rate = 1%; Data 3: latency = 25ms, jitter = 3ms, packet loss rate = 0%; Data 4: latency = 40ms, jitter = 7ms, packet loss rate = 2%; Data 5: latency = 35ms, jitter = 6ms, packet loss rate = 1%; Data 6: delay = 50ms, jitter = 10ms, packet loss rate = 3%; randomly select N data from the collected historical link quality data as the initial centroid of the cluster, N is an integer greater than 1. Here, N=2 can be selected. Correspondingly, the randomly selected initial centroid is the centroid of cluster 1, data 1 (20ms, 2ms, 0%); the centroid of cluster 2, data 4 (40ms, 7ms, 2%). Calculate the distance from each historical link quality data to the centroid of the cluster (Euclidean distance can be used), and assign the data to the nearest cluster. For example, calculate the distance from data 1 to the two centroids: the distance to the centroid of cluster 1 (20, 2, 0) is 0 (because it is the centroid). The distance to the centroid of cluster 2 (40, 7, 2) is:

[0075] Repeat this process, calculate the distances from other data to the two centroids, and finally assign the data to the nearest cluster. Iterate and update the cluster centroids based on the average value of the historical link quality data in the same cluster. Suppose that after the first assignment, data 1, data 2, and data 3 are assigned to cluster 1, and data 4, data 5, and data 6 are assigned to cluster 2. The new centroid 1 of cluster 1 and the new centroid 2 of cluster 2 are:

[0076] Repeat the steps of calculating distance and assigning clusters until the difference between the changes in the centroids of two adjacent updated clusters is less than a preset change threshold (e.g., 0.1). Assume that after multiple iterations, the centroids of the clusters no longer change significantly, and eventually N converged clusters are obtained. According to the centroids of each of the N converged clusters, the convergence regions of each of the N clusters are determined. The convergence region can be defined by setting a radius, which can be based on the average distance from the data points of each cluster to the centroid of the cluster, or based on the standard deviation (the square of the difference between each distance and the average distance, and finally the average of the squared differences is calculated and squared).

[0077] In an optional embodiment, after clustering the link quality data of each of the multiple links to be tested based on the clustering algorithm model obtained by pre-training, the method also includes: determining the cluster center of gravity of each of the multiple clusters, and the link quality indicator corresponding to each cluster center of gravity, wherein the link quality indicator is a one-dimensional indicator; mapping the link quality data of each of the multiple links to be tested to the cluster center of gravity of the cluster to which it belongs, to obtain the link quality indicator of each of the multiple links to be tested, wherein the link quality data includes at least: delay, jitter and packet loss rate.

[0078] The following is a specific example to illustrate how to cluster the link quality data of the link to be tested based on the clustering algorithm model obtained by pre-training, and determine the cluster center of gravity and its corresponding link quality index. Assume that two clusters have been obtained through the clustering algorithm, and the cluster center of gravity of each cluster has been determined. As shown in Table 1, there are now the following three-dimensional link quality data (delay, jitter and packet loss rate) of the link to be tested:

[0079] Table 1 See also Figure 6 ,After clustering processing, the following two clusters and their centroids are obtained: cluster centroid 1 (better performance), (25ms, 3ms, 0.33%); ,cluster centroid 2 (poor performance), (37.5ms, 6.5ms, 1.5%).

[0080] The distance from link 1 to cluster 1 is

[0081] The distance from link 1 to cluster 2 is

[0082] Similarly, calculate the distances of other links to the centroids of the two clusters. According to the calculated distances, select the cluster with the smaller distance as the link's affiliation. It can be obtained that link 1 is closer to cluster 1 and belongs to cluster 1. Link 2 is closer to cluster 1 and belongs to cluster 1. Link 3 is closer to cluster 1 and belongs to cluster 1. Link 4 is closer to cluster 2 and belongs to cluster 2. Link 5 is closer to cluster 2 and belongs to cluster 2.

[0083] Figure 7 is a data distribution diagram of a clustering result in one embodiment of the present disclosure, such as Figure 7 As shown in FIG. 1 , the area included in the dotted circular boundary is the convergence area corresponding to the centroid after convergence. The figure shows five different clusters, and the spatial distribution of the three-dimensional link quality data of each cluster can be seen, for example Figure 7 The quality data of each link in the leftmost cluster represents a higher packet loss rate, while the quality data of each link in the rightmost cluster represents a higher delay.

[0084] In order to map the three-dimensional link quality data of each link into a one-dimensional link quality indicator, a weight can be assigned to each dimension. Assume that the weights of delay, jitter and packet loss rate are . You can define:

[0085] For weighted calculation, the packet loss rate needs to be normalized to obtain the clustering results shown in Table 2:

[0086] Table 2 Figure 8 is a data distribution diagram of another clustering result in one embodiment of the present disclosure, such as Figure 8 As shown, the three-dimensional link quality data is quantified into one-dimensional link quality, which can more intuitively reflect the link quality of each link to be tested. For example, it can be seen that the link corresponding to the link quality data (detection data) in the rightmost cluster represents a worse link quality. The results shown in Table 2 are obtained by weighting the actual link quality data. It can be understood that the higher the value of the link quality data, the higher the link quality index reflected, and the worse the link quality. In other words, the link quality data and link quality are inversely correlated. In order to positively reflect the link quality, the reciprocal processing should be taken in the above-mentioned weighted link quality index results, which can more clearly reflect the link quality. For example, the link quality index of link 1 is the lowest, and the value after taking the reciprocal is the highest. Therefore, compared with the other links, link 1 reflects the best link quality and represents the best link performance.

[0087] The method also includes: monitoring whether there is a faulty link among the multiple links to be tested; if there is a faulty link, determining the cluster center of gravity of the cluster to which the faulty link belongs; from the cluster to which the faulty link belongs, determining a backup link that is closest to the link quality indicator of the abnormal link and whose current remaining bandwidth meets the preset bandwidth requirement; and switching the faulty link to the backup link.

[0088] Continuously monitor the status of all links to be tested. After determining that a link is a faulty link, find the cluster to which the link belongs and determine the center of gravity of the cluster. Specifically, extract the delay, jitter, and packet loss rate of the faulty link. The distance from the faulty link to the center of gravity of each cluster can be calculated using the weighted distance (as described above). Classify the faulty link into the cluster with the closest distance and record the center of gravity of the cluster. Once the cluster of the faulty link and its center of gravity are determined, find a suitable backup link in the cluster. Traverse and check all links belonging to the cluster. All links in the cluster are multiple candidate backup links for the faulty link. Calculate the distance from each link to the link quality indicator of the faulty link. For each candidate backup link, check whether its current remaining bandwidth meets the preset bandwidth requirement. From the links that meet the bandwidth requirement, select the link with the closest distance to the quality indicator of the faulty link as the backup link. Finally, perform link switching to switch the faulty link to the selected backup link.

[0089] Through the above steps, link failures can be effectively monitored, the cluster to which the failed link belongs can be quickly determined, and a suitable backup link can be selected in the cluster for switching, which significantly improves the reliability and stability of the network and ensures business continuity.

[0090] In an optional embodiment, after clustering the link quality data of each of the multiple links to be tested based on the clustering algorithm model obtained by pre-training, the method also includes: collecting multiple historical link quality data of the multiple links to be tested in different time periods to form a historical data set; determining the link quality evolution trend of each of the multiple links to be tested based on the historical data set; based on the clustering results and link quality evolution trend of the link quality data, identifying potential faulty links whose link quality data exceeds the set tolerance range for multiple consecutive time periods, and performing result-oriented analysis on the potential faulty links.

[0091] Through the first device under test, regularly collect the quality data of each link under test. The data should be collected in different time periods to form a time series. Figure 6After collecting enough historical link quality data, combine the clustering results and link quality evolution trends to identify potential faulty links. Specifically, review the clustering results obtained based on the clustering algorithm model to understand the current status of each link. Set the tolerance range of each link quality data based on business requirements and network performance standards. For example, the latency should be less than 50ms, the jitter should be less than 10ms, and the packet loss rate should be less than 1%. Monitor the quality data of each link to determine whether it exceeds the set tolerance range in multiple time periods. If the quality indicator of a link exceeds the tolerance range for multiple consecutive time periods, it is marked as a potential faulty link.

[0092] Once a potential faulty link is identified, an in-depth, result-oriented analysis is performed on it. Specifically, the historical data of the potential faulty link is analyzed in detail to find out the reasons for the degradation of link quality. For example, the changes in latency, jitter, and packet loss rate are analyzed to see whether there are certain time periods or events that cause performance degradation. The relationship between the potential faulty link and other links can also be checked to see if there are common problems, such as network congestion, equipment failure, etc. Based on the analysis results, improvement suggestions for the potential faulty link are proposed, such as adjusting the route, increasing bandwidth, or optimizing the configuration.

[0093] For step S104, Figure 8 The link quality distribution results shown also clearly show that the location of the abnormal detection data point is outside the convergence range, and further analysis can be performed on the abnormal data or the link corresponding to the abnormal data.

[0094] Through the above steps, the link quality evolution trend can be effectively analyzed and potential faulty links can be identified. This method can not only help network administrators find problems in a timely manner, but also provide data support for subsequent troubleshooting and performance optimization, thereby improving network reliability and performance.

[0095] Those skilled in the art will appreciate that the embodiments of the present disclosure may be provided as methods, devices, electronic devices, and storage media. Therefore, the embodiments of the present disclosure may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the embodiments of the present disclosure may take the form of a computer program product implemented on one or more computer-readable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0096] Although the preferred embodiments of the present disclosure have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present disclosure.

[0097] Finally, it should be noted that, in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the term "include" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or terminal device. In the absence of further restrictions, the elements defined by the sentence "include..." do not exclude the existence of other identical elements in the process, method, article or terminal device including the elements. The above is a detailed introduction to a multi-link testing method and system based on SD-WAN provided by the present disclosure. The present disclosure is explained in detail using specific examples in this article to illustrate the principles and implementation methods of the present disclosure. The description of the above embodiments is only used to help understand the method and its core idea of ​​the present disclosure; at the same time, for those skilled in the art, according to the idea of ​​the present disclosure, there will be changes in the specific implementation method and application scope. In summary, the content of this specification should not be understood as a limitation on the present disclosure.

Claims

1. A multi-link testing method based on SD-WAN, characterized in that: The method comprises: Sending a detection message to a second device under test through a first device under test, and receiving a feedback message returned by the second device under test in response to the detection message, wherein the first device under test and the second device under test are respectively configured with a plurality of links under test; Determining link quality data of each of the multiple links to be tested according to the detection message and the feedback message; Based on the clustering algorithm model obtained by pre-training, clustering processing is performed on the link quality data of each of the multiple links to be tested to obtain multiple clusters for characterizing the link quality and convergence regions of each of the multiple clusters; Link quality data that does not belong to the convergence area is determined as abnormal data, and an abnormality analysis is performed on the abnormal data.

2. The method according to claim 1, characterized in that After receiving the feedback message returned by the second device under test in response to the detection message, the method further includes: Determining, according to the detection message and the feedback message, an actual bandwidth of each of the multiple links to be tested; For any link to be tested, comparing the actual bandwidth of the link to be tested with a preset first threshold; When the actual bandwidth is greater than or equal to a preset first threshold, sending a data stream to the first device under test for the link under test through a NETPERF stream-breaking tool of the second device under test; The current remaining bandwidth of the link to be tested is determined by using the SAR traffic monitoring tool of the first device to be tested, and the current remaining bandwidth is used as the link quality data of the link to be tested.

3. The method according to claim 2, characterized in that The method further comprises: Determine the CPU occupancy rate and memory usage rate of each of the multiple links to be tested through the SAR traffic monitoring tool of the first device to be tested; From the multiple links to be tested, determine a CPU-exclusive link to be tested whose CPU occupancy rate is higher than a preset second threshold, and determine a memory leak link to be tested whose memory usage rate is higher than a preset third threshold; For a link under test that is exclusively used by the CPU or a link under test that has a memory leak, the scheduling priority and resource allocation strategy of the link under test are correspondingly adjusted in the QOS yang files of the first device under test and the second device under test.

4. The method according to claim 1, characterized in that: The method further comprises: In the configuration files of the first device under test and the second device under test, respectively, the link uuid number is increased to a preset link upper limit, wherein the link uuid number corresponds one-to-one to the multiple links under test; Monitoring the link quality data of each of the plurality of links under test by using the SAR traffic monitoring tool of the first device under test, wherein the link quality data includes: delay, jitter and packet loss rate; Determining the interface status between the first device under test and the second device under test according to the link quality data of each of the multiple links under test; When the interface status indicates that the interface is blocked, the scheduling priorities and resource allocation strategies of the multiple links to be tested are correspondingly adjusted in the QOS yang files of the first device to be tested and the second device to be tested.

5. The method according to claim 1, characterized in that Follow the steps below to pre-train the clustering algorithm model: Collecting a plurality of historical link quality data of a plurality of links to be tested in different time periods, and inputting the plurality of historical link quality data as training samples into the clustering algorithm model; Randomly selecting N historical link quality data from the plurality of historical link quality data as the first clustering centroid of each of the N clusters, where N is an integer greater than 1; Calculating the distance between each historical link quality data and the centroid of a cluster, and assigning the plurality of historical link quality data to the nearest cluster; Iteratively update the cluster center of gravity according to the average value of the historical link quality data in the same cluster until the change difference between two consecutive updated cluster centers of gravity is less than a preset change threshold, and obtain N converged clusters; According to the cluster centroids of the N converged clusters, the convergence regions of the N clusters are determined.

6. The method according to claim 2, characterized in that After clustering the link quality data of each of the plurality of links to be tested based on the clustering algorithm model obtained through pre-training, the method further includes: Determine the cluster center of gravity of each of the multiple clusters, and a link quality indicator corresponding to each cluster center of gravity, wherein the link quality indicator is a one-dimensional indicator; Mapping the link quality data of each of the multiple links to be tested to the cluster center of gravity of the cluster to which they belong, to obtain the link quality indicators of each of the multiple links to be tested, wherein the link quality data at least includes: delay, jitter and packet loss rate; The method further comprises: Monitoring whether there is a faulty link among the multiple links to be tested; In the case where there is a faulty link, determining the cluster center of gravity of the cluster to which the faulty link belongs; Determine, from the cluster to which the faulty link belongs, a backup link that is closest to the link quality indicator of the abnormal link and whose current remaining bandwidth meets the preset bandwidth requirement; The failed link is switched to the standby link.

7. The method according to claim 1, characterized in that After clustering the link quality data of each of the plurality of links to be tested based on the clustering algorithm model obtained through pre-training, the method further includes: Collect multiple historical link quality data of multiple links to be tested in different time periods to form a historical data set; Determining, according to the historical data set, a link quality evolution trend of each of the multiple links to be tested; Based on the clustering results of the link quality data and the link quality evolution trend, potential faulty links whose link quality data exceeds a set tolerance range for multiple consecutive time periods are identified, and result-oriented analysis is performed on the potential faulty links.

8. A multi-link testing system based on SD-WAN, characterized in that: The system is applied to execute the steps of the method according to any one of claims 1 to 7, the system comprising: a first device under test, a second device under test and a PC device, the first device under test and the second device under test are respectively configured with a plurality of links under test; the first device under test is connected to the PC device via a LAN port, and the first device under test and the second device under test are connected via a WAN port; The first device under test is used to send a detection message to the second device under test, receive a feedback message returned by the second device under test in response to the detection message, and report link quality data of each of the multiple links under test to the PC device according to the detection message and the feedback message; The second device under test is used to receive the detection message and send the feedback message to the first device under test; The PC device performs clustering processing on the link quality data of each of the multiple links to be tested based on the clustering algorithm model obtained through pre-training, and the result of the clustering processing represents the link quality characteristics of each of the multiple links to be tested.

9. The system according to claim 8, characterized in that The first device under test accesses and obtains the IP address of the first device under test through the PC device; the WAN port of the second device under test is set with the same network segment as the first device under test but with a different IP address; The first device under test is added with a first routing detail, and the first routing detail accesses the IP address of the second device under test through the WAN port of the second device under test; The second device under test is added with a second routing detail, and the second routing detail accesses the IP address of the first device under test through the WAN port of the second device under test; The system also includes a WAN port verification module for verifying whether the WAN port of the first device under test can ping the IP address of the WAN port of the second device under test, and verifying whether the WAN port of the second device under test can ping the IP address of the WAN port of the first device under test.

10. The system according to claim 8, characterized in that The first device under test is configured with a first configuration file, and the second device under test is configured with a second configuration file; The first configuration file is used to create the same number of link uuid numbers for the multiple links to be tested, and to set an end detection IP that is the same as the IP address of the second device to be tested, wherein the link uuid number corresponds one-to-one to the multiple links to be tested; The second configuration file is used to create a link uuid number that corresponds one-to-one to the link uuid number configured in the first configuration file, and to set an end detection IP that is the same as the IP address of the first device under test.

Images