Network card stability test method and system

Abstract

The invention provides a network card stability test method and system, and the method comprises the steps: constructing a flow test matrix according to preset flow test parameters, and carrying out the port test parameter extension processing of the flow test matrix, and obtaining a dynamic asymmetric flow test scene set; performing a network card load test according to the dynamic asymmetric flow test scene set, and acquiring network card performance data to obtain a performance index data set; according to the performance index data set, performing multi-stage verification test and signal analysis on the network card to obtain collapse point feature description; and obtaining a network card asymmetric flow stability evaluation report through multi-dimensional data analysis and visualization processing according to the crash point feature description. According to the invention, through construction of a dynamic asymmetric flow test scene, multi-stage verification test and multi-dimensional data analysis, comprehensive test and evaluation of the stability of the network card in a complex asymmetric flow environment are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to a method and system for testing the stability of a network card. Background Art

[0002] With the development of high-performance computing and edge computing, the requirements for network transmission by data acquisition servers are getting higher and higher. As a key hardware component, the stability of the network card directly affects the overall reliability of the system. Existing network card stability testing methods mainly use symmetric traffic patterns or fixed-ratio traffic patterns for testing, and a constant packet size and a linearly increasing load model are used during the testing process to evaluate the performance of the network card. Such traditional testing methods can effectively evaluate the basic performance indicators of the network card, such as throughput, latency, and packet loss rate, in a symmetric traffic environment.

[0003] However, in the actual production environment, especially in data acquisition servers, network cards often face extremely unbalanced two-way traffic (such as the ratio of upstream to downstream traffic reaching 95:5 or more extreme) and rapid fluctuations in traffic ratios. Under such extreme asymmetric traffic conditions, some high-performance network cards may experience sudden performance crashes without triggering alarms. Such failure modes are difficult to detect through traditional symmetric traffic testing methods, resulting in potential stability risks for the system after deployment, which may cause data loss or even system downtime, seriously affecting the reliability of data acquisition servers. Summary of the Invention

[0004] The main objective of the present invention is to solve the technical problem that existing network card stability testing methods cannot effectively detect and evaluate the performance crashes of network cards under extreme asymmetric traffic conditions; In the first aspect of the present invention, a method for testing the stability of a network card is provided. The method for testing the stability of a network card includes: Constructing a traffic test matrix according to preset traffic test parameters, and performing port test parameter expansion processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios; Performing a network card load test according to the set of dynamic asymmetric traffic test scenarios, and collecting network card performance data to obtain a set of performance indicators; Performing multi-level verification testing and signal analysis on the network card according to the set of performance indicators to obtain a description of the characteristics of the crash point; According to the description of the characteristics of the crash point, through multi-dimensional data analysis and visualization processing, obtaining a network card asymmetric traffic stability evaluation report.

[0005] Optionally, in the first implementation manner of the first aspect of the present invention, the constructing a traffic test matrix according to preset traffic test parameters, and performing port test parameter expansion processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios includes: Construct a three-dimensional matrix based on the uplink and downlink traffic ratio parameter, the data packet size combination parameter, and the traffic ratio change speed parameter in the preset traffic test parameters to obtain a traffic test matrix; Calculate the multi-port correlation data based on the multi-port parallel loading factor in the port test parameters for the traffic test matrix; Process the multi-port correlation data according to the mutation parameter in the port test parameters to obtain a traffic mutation point sequence, where the traffic mutation point sequence includes a mutation time point, a mutation amplitude, and a duration; Perform a combined operation on the traffic mutation point sequence and the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set.

[0006] Optionally, in the second implementation manner of the first aspect of the present invention, the performing a combined operation on the traffic mutation point sequence and the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set includes: At each traffic ratio and change speed combination point of the traffic test matrix, calculate a traffic change value according to the mutation time point and the mutation amplitude in the traffic mutation point sequence to obtain traffic superposition data; Perform data point expansion on the traffic superposition data at each mutation duration interval according to a preset sampling interval to obtain a traffic change curve; Generate a single large mutation test sequence, a periodic pulse mutation test sequence, and a random small mutation test sequence respectively according to the traffic change curve to obtain three types of basic test scenarios; Perform a cross combination on the three types of basic test scenarios and the data packet size combination in the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set.

[0007] Optionally, in the third implementation manner of the first aspect of the present invention, the performing a network card load test according to the dynamic asymmetric traffic test scenario set and collecting network card performance data to obtain a performance index data set includes: Configure a central control node and multiple test execution nodes according to the dynamic asymmetric traffic test scenario set to obtain a test topology structure; Perform time synchronization processing on the multiple test execution nodes through the central control node to obtain a synchronized test network; Apply a progressive load under the dynamic perturbation technology to the network card through the synchronized test network to obtain real-time network card response data; Collect network card performance indicators within an overlapping sliding time window according to the real-time network card response data to obtain the performance index data set.

[0008] Optionally, in the fourth implementation manner of the first aspect of the present invention, the multi-level verification test and signal analysis of the network card according to the performance index data set to obtain the crash point feature description includes: Classify the suspected crash points according to the performance degradation amplitude of the performance index data set to obtain classified crash point data; Perform cyclic progressive verification on the complete performance crash points in the classified crash point data to obtain crash point verification data; Perform port cross-verification and peak-valley-peak test on the network card according to the crash point verification data to obtain multi-port verification data; Use a logic analyzer to perform timing analysis on the key signals of the network card in the multi-port verification data to obtain the crash point feature description.

[0009] Optionally, in the fifth implementation manner of the first aspect of the present invention, the performing port cross-verification and peak-valley-peak test on the network card according to the crash point verification data to obtain multi-port verification data includes: Obtain the crash point load value from the crash point verification data, and calculate the first target load value that is 10% higher than the crash point load value to obtain the peak test load value; Apply the peak test load value to the first test port of the network card and maintain it until performance crashes, and at the same time apply a static background traffic with a fixed ratio to other test ports to obtain the first round of cross-verification data; Calculate the second target load value that is 20% lower than the crash point load value, reduce the load of the first test port from the peak test load value to the second target load value, and at the same time apply a fluctuating traffic with a traffic ratio opposite to that of the first test port to other test ports to obtain the second round of cross-verification data; Re-lift the load of the first test port from the second target load value to the peak test load value, and at the same time apply a fluctuating traffic that changes synchronously with the first test port to other test ports to obtain the third round of cross-verification data; Combine the first round of cross-verification data, the second round of cross-verification data, and the third round of cross-verification data into multi-port verification data.

[0010] Optionally, in the sixth implementation manner of the first aspect of the present invention, the obtaining the asymmetric traffic stability evaluation report of the network card through multi-dimensional data analysis and visualization processing according to the crash point feature description includes: Calculate multiple stability dimension indicators according to the crash point feature description, and perform polar coordinate system mapping on the multiple stability dimension indicators to obtain a stability radar chart; Perform stability domain analysis on the two-dimensional plane composed of flow ratio and change rate according to the stability radar chart to obtain the boundary function of the stable operation domain; Perform matching operations on the boundary function of the stable operation domain and different combinations of packet sizes to obtain a network card asymmetric traffic stability evaluation report.

[0011] The second aspect of the present invention provides a network card stability test system, and the network card stability test system includes: A scenario construction module, configured to construct a traffic test matrix according to preset traffic test parameters, and perform port test parameter extension processing on the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set; A load test module, configured to perform network card load testing according to the dynamic asymmetric traffic test scenario set, and collect network card performance data to obtain a performance index data set; A verification and analysis module, configured to perform multi-level verification testing and signal analysis on the network card according to the performance index data set to obtain a description of the characteristics of the crash point; An evaluation and generation module, configured to obtain a network card asymmetric traffic stability evaluation report through multi-dimensional data analysis and visualization processing according to the description of the characteristics of the crash point.

[0012] The above network card stability test method and system, by constructing a traffic test matrix according to preset traffic test parameters, and performing port test parameter extension processing on the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set; performing network card load testing according to the dynamic asymmetric traffic test scenario set, and collecting network card performance data to obtain a performance index data set; performing multi-level verification testing and signal analysis on the network card according to the performance index data set to obtain a description of the characteristics of the crash point; obtaining a network card asymmetric traffic stability evaluation report through multi-dimensional data analysis and visualization processing according to the description of the characteristics of the crash point. The present invention realizes a comprehensive test and evaluation of the stability of the network card in a complex asymmetric traffic environment by constructing a dynamic asymmetric traffic test scenario, multi-level verification testing, and multi-dimensional data analysis.

[0013] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention are realized and obtained by the structures particularly pointed out in the specification, claims, and drawings.

[0014] To make the above objectives, features, and advantages of the present invention more obvious and understandable, the following specifically enumerates preferred embodiments and, in conjunction with the accompanying drawings, makes a detailed description as follows. Description of the Drawings

[0015] Figure 1Schematic diagram of the first embodiment of the network card stability test method in the embodiments of the present invention; Figure 2 Schematic diagram of an embodiment of the network card stability test system in the embodiments of the present invention. Detailed implementation manners

[0016] To make the objectives, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Apparently, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] The terms "including" and "having" and any variations thereof mentioned in the embodiments of the present invention are intended to cover non-exclusive inclusion. For example, a process, method, system, product or device including a series of steps or units is not limited to the listed steps or units, but optionally further includes other steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or devices.

[0018] For ease of understanding of this embodiment, first, a network card stability test method disclosed in the embodiments of the present invention will be introduced in detail. As Figure 1 shown, this method includes the following steps: 101. Construct a traffic test matrix according to preset traffic test parameters, and perform port test parameter extension processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios; In an embodiment of the present invention, the constructing a traffic test matrix according to preset traffic test parameters and performing port test parameter extension processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios includes: constructing a three-dimensional matrix according to the uplink and downlink traffic ratio parameter, packet size combination parameter, and traffic ratio change speed parameter in the preset traffic test parameters to obtain a traffic test matrix; calculating the multi-port correlation data for the traffic test matrix according to the multi-port parallel loading factor in the port test parameters; processing the multi-port correlation data according to the mutation parameter in the port test parameters to obtain a traffic mutation point sequence, where the traffic mutation point sequence includes a mutation time point, a mutation amplitude, and a duration; and performing a combined operation on the traffic mutation point sequence and the traffic test matrix to obtain the set of dynamic asymmetric traffic test scenarios.

[0019] Specifically, the test matrix construction module first obtains preset traffic test parameters, which include the uplink and downlink traffic ratio parameters, the packet size combination parameters, and the traffic ratio change speed parameters. Specifically, the uplink and downlink traffic ratio parameters are set to 9 intervals from 99:1 to 1:99, with a focus on the extreme intervals of 95:5 to 99:1 and 5:95 to 1:99; the packet size combination parameters include 5 mixing ratios of extremely small packets (64 bytes), standard packets (512 bytes), and jumbo packets (9000 bytes); the traffic ratio change speed parameters are set to multiple levels from 1% to 50% change rate per second. Based on these three groups of parameters, a traffic test matrix T(r, p, v) is constructed in a three-dimensional array storage mode, where the r-axis represents the traffic ratio, the p-axis represents the packet size combination, and the v-axis represents the change speed. Each element in this matrix represents a basic test scenario, and the matrix contains a total of 9×5×10 = 450 elements, forming a preliminary test space. This three-dimensional matrix structure enables the test to cover various extreme asymmetric traffic situations, especially the scenario where the uplink data volume is much larger than the downlink data volume commonly found in data acquisition servers. At the same time, it considers the impact of different packet sizes on the network card cache system and processing unit, as well as the challenge of the rapid change of the traffic ratio to the stability of the network card forwarding engine.

[0020] Specifically, the test matrix construction module introduces a multi-port parallel loading factor μ to perform an extended calculation on the traffic test matrix. The value range of the multi-port parallel loading factor μ is from 0 to 1, representing the traffic correlation between multiple physical ports of the network card. When μ = 0, it means that the traffic of each port is completely independent, and when μ = 1, it means that the traffic of all ports is completely synchronized. In actual calculations, five discrete points of μ values {0, 0.25, 0.5, 0.75, 1} are selected, and a transformation function f(T, μ) is applied to each element in the original traffic test matrix to generate multi-port correlation data containing port correlation information. This data is stored in a five-dimensional tensor structure. In addition to the original three dimensions, a port number dimension and a correlation coefficient dimension are added, thus expanding the test space into a more complex test environment that includes multi-port scenarios. The introduction of the multi-port parallel loading factor is a practical consideration for the fact that modern high-performance network cards generally adopt a multi-physical port design. By adjusting the traffic correlation between different ports, the stability of the internal shared resources of the network card when processing multi-port data is tested. The specific implementation of the transformation function f(T, μ) is to apply correlation modulation to the reference traffic of each port. When the μ value increases, the traffic fluctuations of different ports tend to synchronize, which will cause greater pressure on the load distribution unit and forwarding engine inside the network card, thereby comprehensively testing the multi-port processing ability of the network card.

[0021] Specifically, the test matrix construction module further processes the multi-port correlation data according to the mutation parameters to generate a traffic mutation point sequence. The mutation parameters include three aspects: mutation mode type, mutation frequency, and mutation intensity. According to these parameters, a traffic mutation point sequence S composed of multiple triples is generated. Each triple represents the time point of mutation occurrence, mutation amplitude, and duration respectively. In this embodiment, three typical mutation modes are designed: single large mutation (mutation amplitude is 50%-80% of the reference traffic, duration is 5-10 seconds), periodic pulse mutation (period is 30 seconds, mutation amplitude is 20%-40% of the reference traffic, duration is 2-5 seconds), and random small mutation (mutation amplitude is 5%-15% of the reference traffic, duration is 1-3 seconds, occurring randomly). These mutation modes respectively simulate different types of traffic fluctuation situations that the data acquisition server may encounter in actual work. The design of the traffic mutation point sequence directly affects the effectiveness of the test because the network card is more likely to expose potential stability problems when facing mutated traffic. The single large mutation simulates the scenario at the start of large-scale data acquisition, the periodic pulse mutation corresponds to the periodic network load caused by the execution of timed tasks, and the random small mutation simulates the instant communication behavior between distributed nodes. Through the combination of these three typical mutation modes, the traffic mutation point sequence comprehensively covers various traffic change situations that the network card may face in actual work.

[0022] Specifically, the test matrix construction module combines and operates the traffic mutation point sequence with the traffic test matrix to obtain a complete set of dynamic asymmetric traffic test scenarios. The process of the combination operation is to apply each traffic mutation point sequence to each basic test scenario in the traffic test matrix, and generate a dynamically changing traffic pattern through the superposition of time functions. Specifically, for each element T(r, p, v) in the traffic test matrix, combined with each mutation point in the traffic mutation point sequence S, a time-varying traffic modulation is applied to this basic scenario to form a complete test scenario description. The combination operation adopts a specially designed time function mapping algorithm to convert the static traffic configuration into a traffic pattern that changes dynamically over time. The algorithm first divides the test period of a fixed length on the time axis, and then modulates the basic traffic according to the indication of the mutation point sequence within each period. The modulation process takes into account the smoothness of traffic changes and the characteristics of mutations, so that the generated traffic pattern contains both sudden changes and certain continuity, thus being able to more realistically simulate the actual network environment. The finally generated set of dynamic asymmetric traffic test scenarios contains approximately 3000 different test scenarios, and each scenario is a complete time series description, comprehensively covering various working states that the network card may encounter under extreme asymmetric traffic conditions.

[0023] Further, the process of combining the traffic mutation point sequence with the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set includes: at each traffic ratio and change speed combination point of the traffic test matrix, calculating the traffic change value according to the mutation time point and mutation amplitude in the traffic mutation point sequence to obtain traffic superposition data; expanding data points at a preset sampling interval for the traffic superposition data within each mutation duration interval to obtain a traffic change curve; generating a single large mutation test sequence, a periodic pulse mutation test sequence, and a random small mutation test sequence respectively according to the traffic change curve to obtain three types of basic test scenarios; and performing cross-combination of the three types of basic test scenarios with the packet size combination in the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set.

[0024] Specifically, in the process of combining the traffic mutation point sequence with the traffic test matrix, it is first necessary to calculate the traffic change value at each traffic ratio and change speed combination point of the traffic test matrix. Specifically, for each element T(r, v, p) in the traffic test matrix, the system extracts the traffic ratio r (such as 95:5) and the change speed v (such as 10% per second) from it and uses them as the reference traffic configuration. Then, each mutation point in the traffic mutation point sequence S is traversed, and the traffic change value at that moment is calculated according to the mutation time point and mutation amplitude. The calculation method is to multiply the reference traffic by the mutation amplitude. For example, when the reference uplink traffic is 1000 Mbps and the mutation amplitude is 30%, the traffic change value at that moment is 300 Mbps. For each time point, all applicable mutation effects are accumulated to form a complete traffic change time sequence. This step converts the static traffic configuration into a dynamically changing traffic time function, and the generated traffic superposition data is a two-dimensional array containing time points and corresponding traffic increment values. The generation of traffic superposition data takes into account the possible temporal overlap of multiple mutation points and adopts the principle of linear superposition to handle the overlap effect, ensuring that the generated traffic changes conform to the characteristics of the combined action of multiple factors in the actual network environment.

[0025] Specifically, data point expansion processing is performed on the traffic superposition data within each mutation duration interval. Since the mutation point sequence only defines the start time and duration of the mutation, it is necessary to finely describe the traffic changes within the entire duration interval. The system sets a preset sampling interval of 1 millisecond. Within each mutation duration interval, uniform sampling is performed according to this sampling interval to generate a detailed time point sequence. Then, for each sampling time point, the exact traffic value is calculated based on the mutation type and the relative position of the current time within the mutation interval. For example, for the gradually changing mutation type, linear interpolation or S-shaped curve interpolation is used; for the step mutation, the step function is used to describe it. In this way, the sparse mutation point description is extended to a traffic change curve with high time resolution, which contains thousands of data points and accurately describes the process of traffic changing over time. The high time resolution of the traffic change curve ensures that the time characteristics of the traffic change can be precisely controlled during the test, thereby more accurately testing the response ability of the network card to the mutant traffic, especially the processing ability for extremely high traffic situations that change rapidly within a short period of time.

[0026] Specifically, three basic test sequences with different characteristics are generated according to the traffic change curve. First is the single large mutation test sequence, which is characterized by triggering a traffic mutation with an 80% amplitude at a specific time point after the start of the test (usually 30 seconds later), lasting for 10 seconds and then returning to normal. This test simulates the scenario when the data acquisition server starts a large-scale data acquisition task. Second is the periodic pulse mutation test sequence, which has a period of 30 seconds and generates a traffic mutation with a 40% amplitude once in each period, lasting for 5 seconds. This test simulates the periodic impact of timed data synchronization or processing tasks on network traffic. Finally is the random small mutation test sequence, which randomly triggers multiple traffic changes with an amplitude of 10 - 15% during the entire test process, and the duration randomly varies between 1 - 3 seconds. This test simulates the instant communication and small-scale data exchange between distributed nodes. These three types of test sequences respectively focus on testing the adaptability of the network card to different types of traffic changes, thereby comprehensively evaluating the stability performance of the network card in various actual application scenarios.

[0027] Specifically, three types of basic test scenarios are cross - combined with the packet size combinations in the traffic test matrix to generate a final set of dynamic asymmetric traffic test scenarios. The packet size combination dimension includes 5 different configurations, namely pure small packets (64 bytes accounting for 100%), mainly small packets (64 bytes accounting for 80%, 512 bytes accounting for 20%), mixed packets (64 bytes, 512 bytes, and 1500 bytes each accounting for 33%), mainly large packets (1500 bytes accounting for 60%, 9000 bytes accounting for 40%), and pure large packets (9000 bytes accounting for 100%). For each type of basic test sequence, it is combined with these 5 packet size combinations respectively to form 5 variants. In this way, the 3 types of basic test sequences are extended to 15 detailed test scenarios. Then, these 15 test scenarios are applied to each traffic ratio and change speed combination point in the traffic test matrix to form a complete set of test scenarios. The entire set of dynamic asymmetric traffic test scenarios contains approximately 3000 specific test scenarios, each of which consists of two parts: basic parameter configuration (traffic ratio, change speed, packet size combination) and time - series description (traffic change curve), comprehensively covering the working state of the network card under various complex traffic conditions.

[0028] 102. Perform a network card load test according to the set of dynamic asymmetric traffic test scenarios, and collect network card performance data to obtain a performance index data set; In an embodiment of the present invention, the performing a network card load test according to the set of dynamic asymmetric traffic test scenarios and collecting network card performance data to obtain a performance index data set includes: configuring a central control node and multiple test execution nodes according to the set of dynamic asymmetric traffic test scenarios to obtain a test topology structure; performing time synchronization processing on the multiple test execution nodes through the central control node to obtain a synchronized test network; applying a progressive load under the dynamic perturbation technology to the network card through the synchronized test network to obtain real - time response data of the network card; collecting network card performance indicators within an overlapping sliding time window according to the real - time response data of the network card to obtain the performance index data set.

[0029] Specifically, during the network card load test phase, first configure the topology of the test environment according to the dynamic asymmetric traffic test scenario set. Specifically, the system deploys a high-performance server as the central control node C, which is equipped with dual Intel Xeon processors and 128GB of memory, and installs dedicated test control software to be responsible for the coordination and data collection of the entire test process. At the same time, 4 test execution nodes are deployed, and each node is equipped with a 10 Gigabit Ethernet network card and sufficient computing resources to ensure that high-intensity test traffic can be generated. The network card under test is installed on an independent test server and connected to the test execution nodes through a high-speed switch. According to the configuration parameters in the dynamic asymmetric traffic test scenario set, the central control node assigns specific traffic generation tasks to each test execution node, including traffic ratio, packet size distribution, and change patterns, etc. The test topology adopts a star connection method. The central control node is connected to all test execution nodes through an independent management network, while the test execution nodes are connected to the network card under test through a data network. This design of separating the two networks ensures that the test control signaling does not interfere with the actual test traffic and improves the test accuracy. In addition, each test execution node is also configured with dedicated network analysis hardware, such as a high-precision timestamp network card and a hardware traffic generator, to meet the requirements of high-precision testing.

[0030] Specifically, after completing the test topology deployment, the central control node performs precise time synchronization processing on multiple test execution nodes. The time synchronization uses the Precision Time Protocol (PTP, IEEE 1588), which can achieve sub-microsecond-level clock synchronization accuracy on the network. Each test execution node is equipped with a network card supporting PTP and a precise clock source. The central control node, as the master clock, regularly sends synchronization signals and time correction information to all test execution nodes. The time synchronization process first measures the network delay through the PTP transparent transmission function on the switch, then calculates the clock offset, and finally adjusts the clock. The entire time synchronization process continues, and the system performs a synchronization calibration once per second to ensure that the clock deviation of each node during the test always remains within 100 nanoseconds. Through this high-precision time synchronization, all test execution nodes can accurately generate traffic changes strictly according to the predetermined time points, achieving multi-point collaborative traffic control. The establishment of the synchronized test network enables the system to achieve precise coordination of traffic generation and data collection between different test nodes, which is crucial for testing the behavior of the network card when facing complex traffic patterns from multiple ports, especially when testing the competition situation of shared resources inside the network card.

[0031] Specifically, based on the synchronous test network, the system applies a progressive load under dynamic perturbation technology to the network card under test. The progressive load test uses a binary method to detect the performance boundary. First, a coarse-grained scan is performed on the flow ratio dimension, increasing from 50:50 to the extreme ratio (such as 95:5 or 5:95) with a step size of 10%. After finding the interval where the performance begins to decline, the step size is reduced to 2% and continued to detect until the performance breakdown point is accurately located. At each fixed flow ratio, the change speed dimension is detected in a similar way, increasing from low speed (1% per second) to high speed (50% per second). In this process, dynamic perturbation technology is innovatively introduced, that is, high-frequency disturbances with an amplitude of 1-5% of the baseline flow are superimposed on the basic flow pattern, and the disturbance frequency is adjusted in the range of 100Hz-1kHz. For example, when the baseline upstream flow is 900Mbps, a sinusoidal disturbance of ±45Mbps and a frequency of 500Hz is superimposed. This dynamic perturbation can effectively stimulate the potential instability of the network card under extreme conditions, similar to the resonance test of the physical structure, and can discover problems that are difficult to expose in conventional steady-state tests. During the test, the various performance indicators of the network card are monitored in real time. When the performance degradation exceeds the preset threshold (usually 10%), the current test parameters are recorded and the real-time response data of the network card is generated, including time series data such as throughput, latency, packet loss rate, and interrupt frequency.

[0032] Specifically, the system collects NIC performance indicators in overlapping sliding time windows based on the real-time response data of the NIC. The time window analysis method divides the test time axis into a series of overlapping windows, each with a width of 10ms and adjacent windows overlapping by 5ms, forming a continuous analysis view. In each window, the system records 15 key performance indicators: actual throughput, packet delay (minimum, average, maximum, standard deviation), packet loss rate, retransmission rate, traffic ratio, NIC internal queue length (receive queue and send queue), interrupt frequency, DMA transfer completion time, CPU utilization, memory bandwidth utilization, and memory access latency. By modifying the NIC driver, the system can obtain the internal state data of the NIC, especially the operating parameters of the load distribution unit and the internal buffer usage. For each time window, the statistical characteristics of these indicators are calculated, including the mean, standard deviation, maximum, minimum, and percentile distribution. The sliding window method not only provides high-time resolution performance monitoring, but also can capture short-term performance anomalies. All collected performance indicator data are organized into a structured performance indicator data set, which contains test parameters (traffic ratio, change speed, packet size, etc.) and corresponding time series performance data, providing complete experimental evidence for network card stability analysis. This sophisticated data collection method enables the system to detect short-term performance fluctuations and minor performance degradations that are difficult to detect with traditional testing methods, thereby comprehensively evaluating the stability of the network card under extreme asymmetric traffic conditions.

[0033] 103. According to the performance index dataset, perform multi-level verification tests and signal analysis on the network card to obtain a description of the breakdown point characteristics; In an embodiment of the present invention, the performing multi-level verification tests and signal analysis on the network card according to the performance index dataset to obtain a description of the breakdown point characteristics includes: grading the suspected breakdown points according to the performance degradation amplitude of the performance index dataset to obtain graded breakdown point data; performing cyclic progressive verification on the complete performance breakdown points in the graded breakdown point data to obtain breakdown point verification data; performing port cross-verification and peak-valley-peak tests on the network card according to the breakdown point verification data to obtain multi-port verification data; using a logic analyzer to perform timing analysis on the key signals of the network card in the multi-port verification data to obtain a description of the breakdown point characteristics.

[0034] Specifically, the system first accurately grades the suspected breakdown points according to the performance degradation amplitude in the performance index dataset. The grading process uses a data-driven clustering method to divide all recorded performance anomaly points into three levels according to the performance degradation amplitude: slight performance degradation (10%-30%), significant performance decay (30%-70%), and complete performance breakdown (>70%). The calculation of the performance degradation amplitude is based on the comprehensive score of multiple indicators, mainly considering four key indicators: throughput degradation rate, latency increase rate, packet loss rate increase amplitude, and interruption processing time extension ratio. Each indicator is assigned a different weight (throughput 40%, latency 30%, packet loss rate 20%, interruption time 10%) for weighted calculation. For each suspected breakdown point, the system records its complete set of test parameters, including the traffic ratio, change speed, packet size combination, and mutation mode when the breakdown occurs. At the same time, the system also records the detailed performance data 5 seconds before and 10 seconds after the breakdown to form a complete time series description of the breakdown event. After the grading process, the system generates a graded breakdown point database, which contains the complete feature descriptions and detailed classification information of all suspected breakdown points, providing structured data support for the next verification test. The purpose of the grading process is to distinguish performance problems of different severity levels, so as to adopt verification test strategies of different intensities and durations, ensuring both the comprehensiveness of the test and improving the test efficiency.

[0035] Specifically, the system performs cyclic progressive verification on the complete performance breakdown points in the hierarchical breakdown point data to confirm the authenticity and stability of the breakdown points. The cyclic progressive verification adopts a "low-medium-high-medium-low" load change pattern, and the specific execution process is as follows: First, run the system for 30 seconds at a load 10% lower than the breakdown point to establish a performance baseline; then gradually increase the load to the breakdown point at a rate of 1% per second, and record the change curve of the network card performance with the increase of the load; after reaching the breakdown point, maintain this load for 3 minutes to observe whether the network card performance continues to deteriorate or self-recovers; then continue to increase the load to 20% above the breakdown point at a rate of 2% per second to confirm the performance change trend under a higher load; finally, gradually reduce the load back to the initial level and record the process and characteristics of the network card's recovery from the breakdown state. During the entire verification process, the system records a complete set of performance metrics every 100 milliseconds to generate a high-resolution performance-load relationship curve. For each complete performance breakdown point, the system repeats the cyclic progressive verification three times to ensure the consistency and reliability of the results. Through this verification method, the system can not only confirm the exact location of the breakdown point, but also obtain the complete performance change characteristics of the network card before and after approaching the breakdown point, including key characteristics such as the rate of performance decay, the abruptness of the breakdown, and the latency of recovery. The results of the cyclic progressive verification form the breakdown point verification data, which reflect the stability and recovery ability of the network card when facing near-limit asymmetric traffic.

[0036] Specifically, based on the breakdown point verification data, the system further conducts port cross-verification and peak-valley-peak tests on the network card. The core idea of port cross-verification is to apply specific patterns of background traffic to other ports of the network card while confirming the breakdown point of a single port, and observe the mutual influence between multiple ports. When implementing specifically, select one port as the main test port and other ports as background traffic ports, and apply three different patterns of background traffic in sequence: static background traffic (fixed ratio and intensity), reverse fluctuating background traffic (a fluctuating pattern opposite to the traffic ratio of the test port), and synchronous fluctuating background traffic (synchronous fluctuation with the test port). At the same time, perform peak-valley-peak tests on the main test port, that is, first increase the load to 10% above the breakdown point and maintain it until the performance breaks down, then quickly reduce the load to 20% below the breakdown point, record the recovery time and recovery curve, and finally increase the load to the breakdown point again to compare the characteristic differences between the two breakdowns. During the entire test process, the system monitors the performance metrics of all ports and the overall state of the network card, and records the performance correlation and interference patterns between ports. Through this combined test method, the system can evaluate the overall stability of the network card when facing complex multi-port traffic, especially the resource competition situation of internal shared resources (such as caches, processing units) in the network card under multi-port high load. The test results form multi-port verification data, which contain the complete performance characteristics and stability performance of the network card in various multi-port working scenarios.

[0037] Specifically, the system uses a high-precision logic analyzer to perform timing analysis on the key signals of the network card in the multi-port verification data. This step is to analyze the root cause of the network card performance crash from the hardware level. The logic analyzer is configured with a sampling rate of 100MHz and connected to the key test points of the network card through a special test board to collect key hardware signals including interrupt request signals, DMA transfer completion signals, PCIe transaction request signals, internal buffer status signals and clock signals. The acquisition process is carried out simultaneously with the performance test, and high-density sampling is performed in the key time window from 5 milliseconds before the performance crash to 10 milliseconds after the crash. After digital filtering and signal processing, the collected raw signal data is analyzed in detail using professional timing analysis software to identify the timing relationship and abnormal patterns of key events. The analysis focuses on several aspects: the time distribution and pattern of interrupt processing, especially the formation process of interrupt storms; the completion time and failure rate of DMA transfers, reflecting the memory access bottleneck; the filling state changes of the internal buffer, reflecting the data processing capability; the timing relationship between key signals, such as the delay between the interrupt signal and the DMA completion signal. Through these micro-time series characteristics, combined with macro-performance indicators, a comprehensive feature description of the collapse point is systematically constructed, including the triggering conditions, development process, internal mechanisms and manifestation characteristics of the collapse.

[0038] Furthermore, the method of performing port cross-validation and peak-valley-peak testing on the network card according to the crash point verification data to obtain multi-port verification data includes: obtaining a crash point load value from the crash point verification data, and calculating a first target load value that is ten percent higher than the crash point load value to obtain a peak test load value; applying the peak test load value to the first test port of the network card and maintaining it until performance collapses, while applying a fixed ratio of static background traffic to other test ports to obtain a first round of cross-validation data; calculating a second target load value that is twenty percent lower than the crash point load value, reducing the load of the first test port from the peak test load value to the second target load value, while applying a fluctuating traffic with a ratio opposite to that of the first test port to other test ports to obtain a second round of cross-validation data; increasing the load of the first test port from the second target load value back to the peak test load value, while applying a fluctuating traffic that changes synchronously with the first test port to other test ports to obtain a third round of cross-validation data; combining the first round of cross-validation data, the second round of cross-validation data, and the third round of cross-validation data into multi-port verification data.

[0039] Specifically, when performing port cross-validation and peak-valley-peak testing, the system extracts the accurate load value of each confirmed crash point from the crash point verification data. The crash point load value contains parameters in multiple dimensions, mainly the total traffic volume (in Mbps), traffic ratio (such as 95:5), and change speed (such as 20% per second). For each crash point, the system first confirms its stability, that is, the crash point can consistently trigger a performance crash in three repeated validations, with an error within ±3%. Then, the system calculates the first target load value that is 10% higher than the crash point load value as the peak test load value. For example, if the crash point load value of a network card is 9000 Mbps when the uplink-to-downlink traffic ratio is 95:5 and the traffic change speed is 30% per second, the peak test load value is set to 9900 Mbps. When determining the peak test load value, the physical bandwidth limit of the network card needs to be considered. When the calculated result exceeds the physical bandwidth, while keeping the traffic ratio and change speed unchanged, the total volume is set to 99% of the physical bandwidth. For a multi-port network card, the system also needs to specify the main test port (usually the first port) as the application target for peak testing. Determining the peak test load value is crucial for testing the behavior of the network card under overloaded conditions. Setting it 10% higher than the crash point is to ensure that an obvious performance crash can be triggered, while avoiding excessive load causing saturation of the test network itself and affecting the accuracy of the test results.

[0040] Specifically, the system applies the peak test load value to the first test port of the network card to perform the first round of cross-validation. The specific operation is that the central control node instructs the test execution node to send precisely controlled traffic to the first test port, and the traffic parameters are set to the peak test load value calculated in the previous step, including the total volume, ratio, and change speed. At the same time, other test execution nodes apply static background traffic to other ports of the network card. These background traffic has fixed traffic ratios and intensities, usually set to 30% of the maximum bandwidth of the port, and adopt a different traffic ratio from the first port. For example, when the first port has an uplink-to-downlink ratio of 95:5, the background traffic is set to 30:70. During the test process, the system continuously monitors the performance metrics of all ports, including throughput, latency, packet loss rate, queue length, etc., until an obvious performance crash occurs on the first test port (usually defined as a throughput drop of more than 50% or a latency increase of more than 200%). After the performance crash occurs, the system continues to maintain the test load for 5 seconds to observe the performance change trend after the crash. During the entire test process, the system records a complete set of performance metrics every 10 milliseconds to form high-time-resolution performance monitoring data. These data constitute the first round of cross-validation data, reflecting the performance of the network card in the face of the combination of overloading of the main port and static loads on other ports, especially the working status of the load distribution unit and shared cache under this condition.

[0041] Specifically, after the first round of testing causes a performance crash, the system calculates a second target load value that is 20% lower than the load value at the crash point for the second round of cross-validation. Continuing with the previous example, if the load value at the crash point is 9000 Mbps, the second target load value is 7200 Mbps. The system quickly reduces the load on the first test port from the peak test load value (9900 Mbps) to the second target load value (7200 Mbps), while changing the background traffic pattern of other test ports from static background traffic to fluctuating traffic that is opposite to the traffic ratio of the first test port. Specifically, when the traffic ratio of the first test port is 95:5, the traffic ratio of other ports is set to 5:95; when the traffic on the first test port increases, the traffic on other ports decreases accordingly, and vice versa, forming a seesaw traffic fluctuation pattern. This design of reverse fluctuating background traffic aims to test the resource scheduling ability of the network card when processing traffic in the opposite direction. After the load is reduced, the system closely monitors the recovery process of the network card performance, records the time required to recover from the crash state to normal performance, and the performance fluctuation characteristics during the recovery process. This stage of testing continues until the network card performance is completely stable, usually taking 30 - 60 seconds. The performance data collected throughout the process forms the second round of cross-validation data, which reflects the fault recovery ability of the network card and its stability performance when facing traffic interference in the opposite direction.

[0042] Specifically, before the third round of cross-validation begins, the system raises the load on the first test port from the second target load value back to the peak test load value, while applying fluctuating traffic to other test ports that changes synchronously with the first test port. The specific operation is to gradually increase the load on the first test port at a rate of 5% per second until the peak test load value is reached. At the same time, the traffic on other test ports changes in the same pattern and rate as the first test port, but maintains a different traffic ratio. For example, when the traffic on the first test port increases from 7200 Mbps to 9900 Mbps at a rate of 5% per second, the traffic on other ports also increases at the same rate, but the traffic ratio remains opposite to that of the first port (e.g., 95:5 for the first port and 5:95 for other ports). This synchronous fluctuating background traffic pattern tests the overall stability of the network card when facing high-intensity loads with all ports increasing simultaneously, especially the allocation strategy and performance of the internal shared resources of the network card under extreme loads. The testing continues until a performance crash occurs again on the first test port, or the total traffic of all ports reaches the physical bandwidth limit of the network card. The system records the detailed performance data throughout the process to form the third round of cross-validation data, which reflects the performance limit and crash characteristics of the network card under extreme multi-port loads.

[0043] Specifically, the system integrates the cross-validation data of the first three rounds to form a complete multi-port verification data set. The integration process is not just simple data merging but also includes data correlation analysis and feature extraction. First, the system aligns the three rounds of test data in terms of time to ensure the comparability of data from different rounds on the time axis. Then, a series of key feature indicators are calculated, including: crash trigger time (the time from the peak load to the performance crash), recovery time (the time from the load reduction to the performance recovery), crash depth (the maximum amplitude of the performance decline), crash slope (the change rate of the performance from normal to crash), port interference coefficient (the degree of influence of the load of other ports on the performance of the main test port), and re-crash threshold (the ratio of the load value at the second crash to the first one). These feature indicators comprehensively reflect the stability performance of the network card under complex multi-port working conditions. The multi-port verification data set contains not only the original performance measurement data but also these extracted feature indicators.

[0044] 104. According to the description of the crash point features, through multi-dimensional data analysis and visualization processing, a stability evaluation report of the network card's asymmetric traffic is obtained.

[0045] In an embodiment of the present invention, the obtaining of the stability evaluation report of the network card's asymmetric traffic according to the description of the crash point features through multi-dimensional data analysis and visualization processing includes: calculating a plurality of stability dimension indicators according to the description of the crash point features, and performing polar coordinate system mapping on the plurality of stability dimension indicators to obtain a stability radar chart; performing stable domain analysis on the two-dimensional plane composed of the traffic ratio and the change rate according to the stability radar chart to obtain a stable operating domain boundary function; and performing matching operations between the stable operating domain boundary function and different combinations of packet sizes to obtain the stability evaluation report of the network card's asymmetric traffic.

[0046] Specifically, when generating the evaluation report on the stability of the network card's asymmetric traffic, the system calculates multiple stability dimension indicators based on the description of the crash point characteristics to comprehensively characterize the stability characteristics of the network card. Specifically, the system extracts seven key dimension indicators from the description of the crash point characteristics: the critical interval of the traffic ratio R (the most extreme traffic ratio that the network card can stably handle, such as 99:1), the tolerance threshold of the change rate V (the maximum traffic change rate that the network card can stably handle, such as 45% per second), the sensitivity of the packet size P (the degree of influence of different packet sizes on the performance of the network card), the multi-port interference coefficient I (the severity of the mutual influence of the traffic between multiple ports), the recovery time indicator T (the average time required for the network card to recover from the crashed state), the pre-crash characteristic indicator F (the characteristic pattern of the performance degradation before the crash), and the stability margin indicator M (the slope of the performance decay curve near the crash point). Each indicator is calculated from the original test data through a specific algorithm. For example, the multi-port interference coefficient I is calculated by comparing the differences in the crash points between the single-port and the multi-port simultaneous load. All indicators are normalized and converted into values between 0 and 1 for subsequent comprehensive analysis and visual display. These indicators constitute a multi-dimensional representation of the network card's stability, reflecting the performance characteristics and stability boundaries of the network card under extreme asymmetric traffic conditions from different perspectives. Then, the system maps these normalized indicators into a polar coordinate system to generate an intuitive stability radar chart. In the mapping process, the seven dimensions are evenly distributed on the circumference of the polar coordinate, each dimension corresponding to a specific angle, and the indicator value corresponding to the radial distance in that direction. Through this mapping, the stability characteristics of the network card are transformed into a unique closed polygon, the area of which intuitively reflects the overall stability level of the network card, and the shape reflects the distribution of stability in each dimension.

[0047] Next, based on the data from the stability radar chart, the system conducts a stable region analysis on the two-dimensional plane formed by the traffic ratio and the change rate to determine the boundary of the stable operating region of the network card. Specifically, for each packet size combination p, the system plots contour lines on the r-v plane (traffic ratio - change rate plane), representing the combination range of the traffic ratio and the change rate at which the network card can operate stably for that packet size. The boundary of the stable operating region is a closed curve that divides the r-v plane into two regions: the inner region represents the parameter combinations at which the network card can operate stably, and the outer region represents the parameter combinations that will cause performance crashes. The boundary function is derived from the test data points using the piecewise polynomial fitting method to ensure that the curve is smooth and accurately reflects the stability boundary. To enhance the reliability of the boundary, the system conducts additional verification tests at multiple points near the boundary to ensure the accuracy of the boundary position. For each packet size combination, the system generates an independent stable operating region boundary function, and these functions together form a family of stable operating regions of the network card, comprehensively describing the stability characteristics of the network card under different conditions. The stable region analysis pays particular attention to the region where the upstream data flow is dominant and the change rate is high, which is of great significance for evaluating the applicability of the network card in the data acquisition server environment.

[0048] Finally, the system performs a matching operation between the stable operating region boundary function and different packet size combinations to generate a complete asymmetric traffic stability assessment report for the network card. The matching operation first calculates the area of the stable operating region for each packet size combination as a quantitative indicator of the network card's stability under that combination. Then, the system compares the stable operating regions for different packet size combinations to analyze the sensitivity of the network card to the packet size. The assessment report consists of multiple parts: stability summary (including the overall stability score and main features of the network card), multi-dimensional stability radar chart (intuitively showing the stability indicators in seven dimensions), atlas of stable operating regions (showing the stability boundaries under different conditions), packet size sensitivity analysis (analyzing the impact of different packet sizes on stability), and application scenario suitability assessment (specific scores for different application scenarios). The report uses a combination of graphics and text, with intuitive charts and concise written explanations, making it easy for engineers to understand and apply. The assessment report not only provides a comprehensive evaluation of the network card's stability under extreme asymmetric traffic conditions but also provides practical guidance for network card selection and system optimization, especially for data acquisition servers in high-performance computing environments and edge computing deployments. This information is of great reference value for ensuring the stability and reliability of the system.

[0049] In this embodiment, a traffic test matrix is constructed according to preset traffic test parameters, and port test parameter extension processing is performed on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios; a network card load test is performed according to the set of dynamic asymmetric traffic test scenarios, and network card performance data is collected to obtain a performance index data set; according to the performance index data set, multi-level verification testing and signal analysis are performed on the network card to obtain a description of the characteristics of the crash point; according to the description of the characteristics of the crash point, through multi-dimensional data analysis and visualization processing, a network card asymmetric traffic stability evaluation report is obtained. The present invention realizes a comprehensive test and evaluation of the stability of the network card in a complex asymmetric traffic environment by constructing a dynamic asymmetric traffic test scenario, multi-level verification testing, and multi-dimensional data analysis.

[0050] The method for testing the stability of the network card in the embodiment of the present invention has been described above. Next, the system for testing the stability of the network card in the embodiment of the present invention will be described. Please refer to Figure 2 , an embodiment of the system for testing the stability of the network card in the embodiment of the present invention includes: A scenario construction module 201, configured to construct a traffic test matrix according to preset traffic test parameters, and perform port test parameter extension processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios; A load test module 202, configured to perform a network card load test according to the set of dynamic asymmetric traffic test scenarios, and collect network card performance data to obtain a performance index data set; A verification and analysis module 203, configured to perform multi-level verification testing and signal analysis on the network card according to the performance index data set to obtain a description of the characteristics of the crash point; An evaluation generation module 204, configured to obtain a network card asymmetric traffic stability evaluation report through multi-dimensional data analysis and visualization processing according to the description of the characteristics of the crash point.

[0051] In the embodiment of the present invention, the network card stability test system runs the above network card stability test method. The network card stability test system constructs a traffic test matrix according to preset traffic test parameters, and performs port test parameter extension processing on the traffic test matrix to obtain a set of dynamic asymmetric traffic test scenarios; performs a network card load test according to the set of dynamic asymmetric traffic test scenarios, and collects network card performance data to obtain a performance index data set; performs multi-level verification testing and signal analysis on the network card according to the performance index data set to obtain a description of the characteristics of the crash point; according to the description of the characteristics of the crash point, through multi-dimensional data analysis and visualization processing, a network card asymmetric traffic stability evaluation report is obtained. The present invention realizes a comprehensive test and evaluation of the stability of the network card in a complex asymmetric traffic environment by constructing a dynamic asymmetric traffic test scenario, multi-level verification testing, and multi-dimensional data analysis.

[0052] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, devices, or units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be elaborated herein.

[0053] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the essence of the technical solution of the present invention, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to enable a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs.

[0054] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for testing the stability of a network card, characterized in that, The network card stability test method includes: Construct a traffic test matrix according to preset traffic test parameters, and perform port test parameter extension processing on the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set; Perform a network card load test according to the dynamic asymmetric traffic test scenario set, and collect network card performance data to obtain a performance index data set; Perform multi-level verification tests and signal analysis on the network card according to the performance index data set to obtain a description of the breakdown point characteristics; According to the description of the breakdown point characteristics, obtain a network card asymmetric traffic stability evaluation report through multi-dimensional data analysis and visualization processing.

2. The network card stability test method according to claim 1, wherein The step of constructing a traffic test matrix according to preset traffic test parameters, and performing port test parameter extension processing on the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set includes: Construct a three-dimensional matrix according to the uplink and downlink traffic ratio parameter, data packet size combination parameter, and traffic ratio change speed parameter in the preset traffic test parameters to obtain a traffic test matrix; Calculate the multi-port correlation data for the traffic test matrix according to the multi-port parallel loading factor in the port test parameters; Process the multi-port correlation data according to the mutation parameter in the port test parameters to obtain a traffic mutation point sequence, where the traffic mutation point sequence includes a mutation time point, a mutation amplitude, and a duration; Perform a combined operation on the traffic mutation point sequence and the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set.

3. The network card stability test method according to claim 2, wherein The step of performing a combined operation on the traffic mutation point sequence and the traffic test matrix to obtain the dynamic asymmetric traffic test scenario set includes: At each traffic ratio and change speed combination point of the traffic test matrix, calculate the traffic change value according to the mutation time point and mutation amplitude in the traffic mutation point sequence to obtain traffic superposition data; Perform data point expansion on the traffic superposition data at each mutation duration interval according to a preset sampling interval to obtain a traffic change curve; According to the traffic change curve, generate a single large mutation test sequence, a periodic pulse mutation test sequence, and a random small mutation test sequence respectively to obtain three types of basic test scenarios; Perform cross-combination on the three types of basic test scenarios and the data packet size combination in the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set.

4. The network card stability test method according to claim 1, characterized in that, The step of performing a network card load test according to the dynamic asymmetric traffic test scenario set, and collecting network card performance data to obtain a performance index data set includes: Configure a central control node and multiple test execution nodes according to the dynamic asymmetric traffic test scenario set to obtain a test topology structure; Perform time synchronization processing on the multiple test execution nodes through the central control node to obtain a synchronized test network; Apply a progressive load under the dynamic perturbation technology to the network card through the synchronized test network to obtain real-time network card response data; Collect network card performance indicators within an overlapping sliding time window according to the real-time network card response data to obtain the performance index data set.

5. The network card stability test method according to claim 1, characterized in that Based on the performance index data set, perform multi-level verification tests and signal analysis on the network card, and the obtained description of the crash point characteristics includes: Grade the suspected crash points according to the performance degradation amplitude of the performance index data set to obtain graded crash point data; Perform cyclic progressive verification on the complete performance crash points in the graded crash point data to obtain crash point verification data; Perform port cross-verification and peak-valley-peak tests on the network card according to the crash point verification data to obtain multi-port verification data; Use a logic analyzer to perform timing analysis on the key signals of the network card in the multi-port verification data to obtain a description of the crash point characteristics.

6. The network card stability test method according to claim 5, wherein, The step of performing port cross-verification and peak-valley-peak tests on the network card according to the crash point verification data to obtain multi-port verification data includes: Obtain the crash point load value from the crash point verification data, and calculate the first target load value that is 10% higher than the crash point load value to obtain the peak test load value; Apply the peak test load value to the first test port of the network card and maintain it until performance crashes, and at the same time apply a static background traffic with a fixed ratio to other test ports to obtain the first-round cross-verification data; Calculate the second target load value that is 20% lower than the crash point load value, reduce the load of the first test port from the peak test load value to the second target load value, and at the same time apply a fluctuating traffic with a flow ratio opposite to that of the first test port to other test ports to obtain the second-round cross-verification data; Raise the load of the first test port from the second target load value back to the peak test load value, and at the same time apply a fluctuating traffic that changes synchronously with the first test port to other test ports to obtain the third-round cross-verification data; Combine the first-round cross-verification data, the second-round cross-verification data, and the third-round cross-verification data into multi-port verification data.

7. The network card stability test method according to claim 1, characterized in that The step of obtaining the network card asymmetric traffic stability evaluation report through multi-dimensional data analysis and visualization processing according to the crash point characteristics description includes: Calculate multiple stability dimension indicators according to the crash point characteristics description, and perform polar coordinate system mapping on the multiple stability dimension indicators to obtain a stability radar chart; Perform stable domain analysis on the two-dimensional plane composed of the traffic ratio and the change rate according to the stability radar chart to obtain the stable operation domain boundary function; Match the stable operation domain boundary function with different packet size combinations to obtain the network card asymmetric traffic stability evaluation report.

8. A network card stability testing system, characterized in that The network card stability test system includes: A scenario construction module for constructing a traffic test matrix according to preset traffic test parameters and performing port test parameter expansion processing on the traffic test matrix to obtain a dynamic asymmetric traffic test scenario set; A load test module for performing network card load tests according to the dynamic asymmetric traffic test scenario set and collecting network card performance data to obtain a performance index data set; A verification and analysis module for performing multi-level verification tests and signal analysis on the network card according to the performance index data set to obtain a description of the crash point characteristics; An evaluation generation module, configured to obtain a stability evaluation report on the asymmetric traffic of the network card through multi-dimensional data analysis and visualization processing according to the described crash point characteristics.

Images