Multi-thread service concurrent request performance evaluation method, system and equipment and storage medium

Abstract

The invention provides a multi-thread service concurrent request performance evaluation method, system and device and a storage medium, and relates to the technical field of network performance evaluation by traversing multi-dimensional resource configuration parameters to determine an abnormal test rate and a non-abnormal throughput rate under combined configuration of each single thread count and single request concurrent number to obtain a throughput stability loss rate. According to the method, the stability critical curve is constructed, the construction of the stability critical curve provides a visual basis for resource allocation decision, the influence of the thread count and the request concurrency number on the performance is accurately quantified through composite index analysis, the dependence on artificial experience judgment is reduced, and the accuracy of resource allocation decision making is improved. The problem that index correlation analysis is insufficient in a traditional evaluation method is effectively solved, the performance bottleneck is accurately positioned until a resource allocation strategy is optimized, and the bearing capacity and stability of a server in a high-concurrency scene are improved. The objectivity and accuracy of performance evaluation are improved, and the technical effect of achieving optimal balance between the resource utilization rate and the performance stability is achieved.

Description

Technical Field

[0001] This application relates to the technical field of network performance evaluation, and particularly to a method, system, device, and storage medium for evaluating the concurrent request performance of multi-threaded services. Background Art

[0002] Emerging scenarios such as the Internet of Things and the Data Internet of Things have given rise to a huge number of high-concurrency demands. Accurately evaluating the bearing capacity of a server in a real scenario can guide the optimal allocation of resources and avoid the decline of user experience or the sharp increase of operation and maintenance costs caused by performance bottlenecks.

[0003] In the prior art, performance evaluation tools are usually used to simulate concurrent requests and generate isolated index reports, lacking the analysis of index relevance, making it difficult to quantify the performance boundary and effectively guide the optimization direction. Summary of the Invention

[0004] This application provides a method, system, device, and storage medium for evaluating the concurrent request performance of multi-threaded services to solve the problem in the prior art that it is impossible to conduct index relevance analysis based on test results and guide the optimization direction.

[0005] To solve the above problems, this application discloses a method for evaluating the concurrent request performance of multi-threaded services, including the following steps: Traverse each single-thread number in the thread number set and each single-request concurrency number in the request concurrency number set in sequence to obtain the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number configuration. The abnormal test rate represents the proportion of the number of tests in which the key performance evaluation index exceeds the preset threshold to the total number of tests, and the non-abnormal throughput rate represents the statistical value of the throughput rate in which the key performance evaluation index does not exceed the preset threshold; Perform a linear operation on the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number to obtain the throughput stability loss rate. According to the throughput stability loss rate, construct a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate. The stable critical curve represents the critical point of the balance relationship between the concurrent request performance and stability of the multi-threaded service under each single-thread number and each single-request concurrency number configuration.

[0006] To solve the above problems, this application also discloses a system for evaluating the concurrent request performance of multi-threaded services, which is applied to the method for evaluating the concurrent request performance of multi-threaded services. The system includes: A system control module is used to sequentially traverse each single-thread number in the set of thread numbers and each single-request concurrency number in the set of request concurrency numbers, so as to obtain the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number configuration. The abnormal test rate represents the proportion of the number of tests in which the key performance evaluation index exceeds the preset threshold to the total number of tests, and the non-abnormal throughput rate represents the statistical value of the throughput rate in which the key performance evaluation index does not exceed the preset threshold. A statistical estimation module is used to perform a linear operation on the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number to obtain a throughput stability loss rate. According to the throughput stability loss rate, a stable critical curve is constructed with the single-request concurrency number as the abscissa and the single-thread number as the ordinate. The stable critical curve represents the critical point of the balance relationship between the multi-threaded service concurrent request performance and stability under each single-thread number and each single-request concurrency number configuration.

[0007] To solve the above problems, the present application also discloses an electronic device, including: A processor; A memory for storing executable instructions of the processor; Wherein, the processor is configured to execute the instructions to implement the method for evaluating the concurrent request performance of the thread service.

[0008] To solve the above problems, the present application also discloses a computer-readable storage medium. When the instructions in the computer-readable storage medium are executed by the processor of the terminal, the terminal can execute the method for evaluating the concurrent request performance of the thread service.

[0009] Compared with the prior art, the present application has the following advantages: By traversing multi-dimensional resource configuration parameters, determining the abnormal test rate and the non-abnormal throughput rate under each combination configuration of single-thread number and single-request concurrency number, performing a linear operation on the abnormal test rate and the non-abnormal throughput rate to obtain a throughput stability loss rate, and constructing a stable critical curve, the construction of the stable critical curve provides an intuitive visual basis for resource configuration decision-making. Through composite index analysis, it accurately quantifies the impact of the number of threads and request concurrency on performance, reduces the dependence on manual experience judgment, effectively solves the problem of insufficient index correlation analysis in traditional evaluation methods, realizes accurate positioning of performance bottlenecks, until optimizing the resource allocation strategy, and improves the bearing capacity and stability of the server in high-concurrency scenarios. It improves the objectivity and accuracy of performance evaluation, and achieves the technical effect of achieving the optimal balance between resource utilization rate and performance stability. Description of the Drawings

[0010] Figure 1Shows the method flowchart of the existing test tool for server performance evaluation test; Figure 2 Shows the method flowchart of a multi-threaded service concurrent request performance evaluation method provided by an embodiment of the present application; Figure 3 Shows the heat map of the throughput stability loss rate provided by an embodiment of the present application; Figure 4 Shows the schematic diagram of the stable critical curve provided by an embodiment of the present application; Figure 5 Shows the method flowchart of resource optimization configuration according to the stable critical curve provided by an embodiment of the present application; Figure 6 Shows the schematic diagram of the process for determining the abnormal test and the throughput rate without abnormality provided by an embodiment of the present application; Figure 7 Shows the example diagram of the method process of the concurrent number of total requests per single time provided by an embodiment of the present application; Figure 8 Shows the statistical chart of the throughput rate change provided by an embodiment of the present application; Figure 9 Shows the method flowchart of determining the abnormal test provided by an embodiment of the present application; Figure 10 Shows the method flowchart of determining the abnormal test provided by another embodiment of the present application; Figure 11 Shows the example diagram of the method process for determining the abnormal test provided by an embodiment of the present application; Figure 12 Shows the schematic diagram of the process for determining the abnormal test and the throughput rate without abnormality provided by an embodiment of the present application; Figure 13 Shows the method flowchart of drawing the stable critical curve provided by an embodiment of the present application; Figure 14 Shows the method flowchart of determining the abnormal test rate and the throughput rate without abnormality provided by an embodiment of the present application; Figure 15 Shows the schematic diagram of the structure of a multi-threaded service concurrent request performance evaluation system provided by an embodiment of the present application; Figure 16 Shows the schematic diagram of the structure of a multi-threaded service concurrent request performance evaluation system provided by another embodiment of the present application; Figure 17 Shows the schematic diagram of the structure of an electronic device provided by an embodiment of the present application. Detailed implementation manners

[0011] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0012] In the evolution process of the Internet and World Wide Web services, the mature B / S architecture protocol system effectively supports the development of business functions and the needs of large-scale user access. However, with the in-depth development of Internet of Things and Data Internet technologies, vertical domain protocols and services have shown explosive growth, and the complexity of their business scenarios and the scale of concurrency far exceed the traditional scope. This poses higher requirements for the performance evaluation of network services.

[0013] In the performance evaluation of network services, core performance indicators such as latency, throughput, and concurrency are usually evaluated. Among them, the maximum processing capacity of the server, namely QPS (Queries Per Second) and TPS (Transactions Per Second), are key indicators to measure server performance. In the prior art, common server performance evaluation and testing tools include Apache Bench, Siege, and JMeter. These tools can complete single or multiple performance tests and provide corresponding measurement results and statistical reports. Referring to Figure 1 , a method flowchart of using existing testing tools for server performance evaluation and testing is shown. After setting the number of threads and the request concurrency, the test is started to obtain performance indicators such as throughput, average response time, test time, and standard deviation of response time.

[0014] In the process of implementing the inventive technical solution in the embodiments of the present application, the inventors found that the above technologies have at least the following technical problems: Lack of analysis of index correlation: In the prior art, during the performance evaluation process, each performance index is presented independently. For example, key indicators such as throughput and the number of threads are usually only presented side by side as the results of single tests, without revealing the internal correlation and interaction mechanism between the two indicators.

[0015] Unclear test conditions and performance boundaries: Existing tools cannot accurately quantify the dependence relationship between performance indicators and system resources. Although they can provide independent performance data under specific test conditions, they fail to clearly show how the configuration of computing resources and network resources restricts performance indicators and cannot provide quantitative analysis data with guiding significance and performance change trends.

[0016] Insufficient presentation of performance change trends and limit values: Under the constraint of computing resources, traditional tools cannot determine the number of CPUs required by the service and the optimal thread configuration. The prior art usually only presents static test results and cannot reflect the dynamic trends of performance indicators with changes in resource configuration and performance limits.

[0017] Insufficient intelligence in data analysis: The performance data analysis of existing technologies relies heavily on the experience and judgment of technicians. The simple statistical charts provided by traditional tools lack the ability to locate problems and cannot directly and objectively reveal the performance bottlenecks of the system. Instead, it depends on the subjective speculation and analysis of technicians.

[0018] In order to solve the significant limitations of existing testing tools in simulating high-dimensional loads, analyzing associated performance metrics, and quantifying resource constraint boundaries, which can no longer meet the accuracy, efficiency, and reliability requirements of new-form services for performance evaluation. This application proposes a method, system, device, and storage medium for evaluating the concurrent request performance of multi-threaded services.

[0019] Figure 2 Shows a flowchart of a method for evaluating the concurrent request performance of multi-threaded services provided by an embodiment of this application. Referring to Figure 2 , an embodiment of this application provides a method for evaluating the concurrent request performance of multi-threaded services, including the following steps: S10. Traverse each single-thread count in the thread count set and each single-request concurrency count in the request concurrency count set in sequence to obtain the abnormal test rate and the non-abnormal throughput rate under each configuration of single-thread count and single-request concurrency count. The abnormal test rate represents the proportion of the number of tests in which the key performance evaluation indicators exceed the preset threshold to the total number of tests, and the non-abnormal throughput rate represents the statistical value of the throughput rate in which the key performance evaluation indicators do not exceed the preset threshold.

[0020] Specifically, the thread count set refers to a set containing different thread values, the request concurrency count set refers to a set containing different request concurrency values, the single-thread count refers to a thread value used for performance evaluation in the thread count set, the single-request concurrency count refers to a request concurrency value used for performance evaluation in the request concurrency count set, the abnormal test rate refers to the proportion of the number of tests in which the key performance evaluation indicators exceed the preset threshold, and the non-abnormal throughput rate refers to the statistical value of the throughput rate that does not exceed the preset threshold.

[0021] Exemplarily, the set of thread numbers can be implemented by dynamically generating thread values in combination with the number of server CPUs and evaluation requirements to cover thread configurations under different computing resources and solve the problem of unclear test conditions and performance boundaries in the prior art; the set of request concurrency numbers can be implemented by generating discrete values with a continuous integer interval and a preset sampling interval to simulate the gradient change of a large number of concurrent requests in a real scenario and solve the problem that traditional tools cannot present the performance change trend; a test with a key performance evaluation indicator exceeding a preset threshold is an abnormal test, and a test with a key performance evaluation indicator not exceeding the preset threshold is a non-abnormal test. The abnormal test rate is determined by calculating the ratio of the number of abnormal tests to the total number of tests under a single configuration of the number of threads and the number of concurrent requests; the non-abnormal throughput rate is calculated by taking the arithmetic mean or weighted mean of the throughput rates of non-abnormal tests.

[0022] In this embodiment, first, each single-thread number in the set of thread numbers and each single-request concurrency number in the set of request concurrency numbers are traversed in sequence. For each combination configuration of the number of threads and the number of request concurrencies, a performance evaluation test is performed and the test results are recorded. The test results at least include key performance evaluation indicators. By analyzing the key performance evaluation indicators, two key indicators, namely the abnormal test rate and the non-abnormal throughput rate, are calculated.

[0023] S20. Perform a linear operation on the abnormal test rate and the non-abnormal throughput rate for each single-thread number and each single-request concurrency number to obtain a throughput stability loss rate. According to the throughput stability loss rate, construct a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate. The stable critical curve represents the critical point of the balance relationship between the multi-threaded service concurrent request performance and stability under each configuration of the single-thread number and the single-request concurrency number.

[0024] Specifically, the throughput stability loss rate refers to a linear combination index of the abnormal test rate and the non-abnormal throughput rate, which can be specifically implemented by setting weight coefficients and performing weighted summation on the two indicators. The stable critical curve refers to a curve representing the balance relationship between performance and stability, which can be specifically implemented by using heat map data fitting and a segmentation algorithm to draw the boundary line. The stable critical curve visually displays the performance limit values under different resource configurations, solving the technical problem that traditional tools cannot guide resource optimization configuration.

[0025] Exemplarily, Table 1 shows a throughput stability loss rate statistical table provided by an embodiment of the present application. Referring to Table 1, Ty represents the single-thread number, and Cx represents the single-request concurrency number. Table 1 shows the throughput stability loss rates under each configuration of the single-request number and the number of threads.

[0026] Table 1

[0027] Exemplarily, Figure 3 a heat map of the throughput stability loss rate provided by an embodiment of the present application is shown, Figure 4 and a schematic diagram of the stable critical curve provided by an embodiment of the present application is shown. Referring to Figure 3 and Figure 4 , after determining the throughput stability loss rate under each single-thread number and each single-request concurrency number, first, according to the data shown in Table 1, a heat map is constructed with the single concurrency number as the abscissa, the single-thread number as the ordinate, and the throughput stability loss rate as the color scale, as shown in Figure 3 . Then, through a preset fitting and segmentation algorithm, the stable critical curve is drawn.

[0028] In the embodiment of the present application, first, each single-thread number in the thread number set and each single-request concurrency number in the request concurrency number set are traversed in sequence. For each combination configuration of the single-thread number and the single-request concurrency number, a performance evaluation test is performed and the test result is recorded. By analyzing the key performance evaluation indicators in the test result, the abnormal test rate and the non-abnormal throughput rate under each combination configuration of the single-thread number and the single-request concurrency number are determined; then, through a linear operation on the abnormal test rate and the non-abnormal throughput rate under each combination configuration of the single-thread number and the single-request concurrency number, the throughput stability loss rate is obtained, and a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate is constructed, which intuitively reflects the scoring relationship critical point between the multi-thread service concurrent request performance and stability under different single-thread number and single-request concurrency number configurations.

[0029] By traversing the multi-dimensional resource configuration parameters, the abnormal test rate and the non-abnormal throughput rate under each combination configuration of the single-thread number and the single-request concurrency number are determined, a linear operation is performed on the abnormal test rate and the non-abnormal throughput rate to obtain the throughput stability loss rate, and the stable critical curve is constructed. The construction of the stable critical curve provides an intuitive visual basis for resource configuration decision-making. Through composite index analysis, the influence of the thread number and the request concurrency number on performance is accurately quantified, the dependence on manual experience judgment is reduced, the problem of insufficient index correlation analysis in the traditional evaluation method is effectively solved, the performance bottleneck is accurately located, until the resource allocation strategy is optimized, and the bearing capacity and stability of the server in the high-concurrency scenario are improved. The objectivity and accuracy of performance evaluation are improved, and the technical effect of achieving the optimal balance between resource utilization rate and performance stability is realized.

[0030] Figure 5 A method flow chart for resource optimization configuration according to the stable critical curve provided by an embodiment of the present application is shown. Referring to Figure 5 , the multi-thread service concurrent request performance evaluation method further includes the following steps: S30. When the preset target concurrency is determined, determine the single-thread number corresponding to the stable critical curve as the minimum thread configuration.

[0031] S40. When the preset target thread number is determined, determine the single-request concurrency corresponding to the stable critical curve as the maximum request concurrency.

[0032] Specifically, the preset target concurrency refers to the preset request concurrency, which is used to determine the minimum thread configuration that meets the performance requirements; the preset target thread number refers to the total number of parallel processing threads preset by the server, which is used to evaluate the maximum request concurrency threshold that can be tolerated under the preset target thread number configuration.

[0033] Exemplarily, when the preset target concurrency is determined, through the preset target concurrency and the stable critical curve, the corresponding single-thread number can be determined, and through the corresponding single-thread number, the minimum thread configuration under the preset target concurrency can be further determined; when the preset target thread number is determined, through the preset target thread number and the stable critical curve, the corresponding single-request concurrency can be determined, and further determine the maximum request concurrency that can be tolerated under the preset target thread number configuration. When determining the minimum thread configuration, the ceiling strategy is adopted. When the single-thread number is a non-integer, it is rounded up to the adjacent integer, and the minimum thread configuration is determined according to the closest higher integer.

[0034] For example, referring to Figure 4 and Figure 5 , when the preset target concurrency is 200, the minimum thread configuration is 4, that is, when the thread configuration is greater than or equal to 4, the performance requirements can be met; when the preset target concurrency is 250, the minimum thread configuration is 5, that is, when the thread configuration is greater than or equal to 5, the performance requirements can be met; when the preset target thread number is 4, the maximum request concurrency is 225, that is, the maximum request concurrency that can be tolerated is 225; when the preset target thread number is 3, the maximum request concurrency is 125, that is, the maximum request concurrency that can be tolerated is 125.

[0035] In the embodiments of the present application, through the stable critical curve for resource optimization configuration, the minimum thread configuration can be accurately determined according to the preset target concurrency, or the maximum request concurrency can be evaluated under the preset target thread number. The ceiling strategy is adopted to ensure the practicability of resource optimization configuration, effectively solving the problem of unreasonable resource configuration in the traditional method in the dynamic concurrency scenario, and significantly improving the performance stability and resource utilization rate of the multi-thread service in the high-concurrency environment.

[0036] Figure 6 shows a schematic flowchart of the process for determining the abnormal test and the throughput rate without anomalies provided by an embodiment of the present application. Referring toFigure 6 , S10: Obtain the exception test rate and the throughput rate without exceptions under each single-thread count and each single-request concurrency count configuration, including the following steps: S11A: Starting from the initial total request concurrency count, increase it step by step according to a preset step size to obtain the total request concurrency count. Under each total request concurrency count, record the throughput rate of the single-thread count and single-request concurrency count configuration. When the fluctuation range of the throughput rate is less than the fluctuation threshold, determine the minimum value among all the total request concurrency counts corresponding to the maximum throughput rate, and generate the single total request concurrency count.

[0037] Specifically, the total request concurrency count covers different load scenarios by increasing step by step. The preset step size can be set as a fixed value or a dynamically adjusted value according to the evaluation requirements. The fluctuation range of the throughput rate is quantified by calculating the variance or standard deviation of the throughput rate. The total request concurrency count starts from the initial total request concurrency count and gradually increases. After each increase of the preset step size, perform a test and record the throughput rate. When the fluctuation range of the throughput rate is less than the fluctuation threshold, select the minimum total request concurrency count corresponding to the throughput rate branch as the minimum total request concurrency count, which can avoid abnormal performance indicators caused by too high a concurrency count.

[0038] Exemplarily, Figure 7 shows a flowchart example of the method for the single total request concurrency count provided by an embodiment of the present application, Figure 8 shows a statistical chart of the throughput rate change provided by an embodiment of the present application. Refer to Figure 7 , set the initial total request concurrency count N = N0 and the step size Nstep, and at the same time initialize the intermediate value Nout = 0. First, test the throughput rate under the single-thread count Cx, single-request concurrency count Ty, and the initial total request concurrency count. Then, use the updated total request count N = N + Nstep, and test the throughput rate under the single-thread count Cx and single-request concurrency count Ty again; when the throughput rate increases, record Nout as the current N value. When the throughput rate does not increase, update the current Nout0 as the current N value, and determine whether Nout0 is greater than Nout. When Nout0 is not greater than Nout, continue to increase the total request concurrency count for testing until the fluctuation range of the throughput rate is less than the preset threshold, and determine that the stable determination count is reached. When Nout0 is greater than Nout, take the maximum value of Nout0 and Nout as the new Nout, and continue to test the throughput rate. When the stable determination count is reached, record Nout and determine the single total request concurrency count.

[0039] Refer to Figure 8, when the number of single-threaded threads is 1, the number of concurrent requests per single request is 10, the preset step size is 100, and the fluctuation threshold is 1%, as the total number of concurrent requests increases, it can be seen that when the fluctuation range of the throughput rate is less than the fluctuation threshold, the total number of concurrent requests is between 1400 and 2000, and the single total number of concurrent requests is determined to be 1400.

[0040] S12A. Perform N performance evaluation tests according to the number of single-threaded threads, the number of concurrent requests per single request, and the single total number of concurrent requests, and obtain N test results. Each test result includes at least the throughput rate and the key performance evaluation indicators.

[0041] S13A. When the key performance evaluation indicators exceed the preset threshold, determine the performance evaluation test to which the key performance evaluation indicators belong as an abnormal test, calculate the proportion of the number of abnormal tests in the total number of N performance evaluation tests to obtain the abnormal test rate, and determine the non-abnormal throughput rate according to the throughput rate of the non-abnormal tests in the N performance evaluation tests.

[0042] Specifically, the calculation of the abnormal test rate is based on statistical proportion, and the influence of random errors is reduced through N independent tests. The value of N can be set according to the confidence requirement; the non-abnormal throughput rate aggregates the throughput rate data of non-abnormal tests by arithmetic mean or weighted mean to ensure the representativeness of the obtained non-abnormal throughput rate.

[0043] In the embodiments of the present application, through the combination of the preset step size and the fluctuation threshold, it is ensured that the test process covers the state change of the server from low load to critical load, effectively determines the single total number of concurrent requests, accurately reflects the performance boundary of the server in the high-concurrency scenario, clarifies the processing ability of the server in the stable state, and provides a reasonable load benchmark for the subsequent performance evaluation test by determining the single total number of concurrent requests, making the test results more valuable for reference and ensuring the accuracy and stability of the performance evaluation. By introducing the abnormal test rate and the non-abnormal throughput rate, it is possible to comprehensively reflect the performance of the server under high load, avoid one-sided evaluation that may be caused by relying on a single indicator, and help better understand the performance characteristics of the system.

[0044] Exemplarily, the key performance evaluation indicators at least include the test time, and the preset threshold at least includes the first threshold.

[0045] Figure 9 Shows a flowchart of a method for determining an abnormal test provided by an embodiment of the present application. Refer to Figure 9 , S13A. When the key performance evaluation indicators exceed the preset threshold, determine the performance evaluation test to which the key performance evaluation indicators belong as an abnormal test, including the following steps: S131a. Set the first threshold according to the single total number of concurrent requests.

[0046] S132a. When the test time exceeds the first threshold, determine the performance evaluation test to which the key performance evaluation indicator belongs as an abnormal test.

[0047] Specifically, the test time refers to the actual total time consumed from the start of the test to the processing of all requests for the concurrent number of total requests per single time, and the first threshold refers to the theoretical maximum total time consumed from the start of the test to the processing of all requests for the concurrent number of total requests per single time. The larger the concurrent number of total requests per single time, the larger the first threshold. By determining the first threshold according to the concurrent number of total requests per single time and triggering the abnormal marking mechanism when the test time exceeds the first threshold to determine the performance evaluation test as an abnormal test, abnormal tests can be effectively identified and screened. By setting the first threshold related to the concurrent number of total requests per single time, it is ensured that the determination conditions for abnormal tests can adapt to the test scenarios under different concurrent numbers of total requests per single time, improving the applicability and accuracy of performance evaluation.

[0048] Exemplarily, the key performance evaluation indicators at least include the standard deviation of the request response time, and the preset thresholds at least include the second threshold.

[0049] Figure 10 Shows the flowchart of the method for determining abnormal tests provided by another embodiment of the present application. Refer to Figure 10 , S13A. When the key performance evaluation indicator exceeds the preset threshold, determine the performance evaluation test to which the key performance evaluation indicator belongs as an abnormal test, including the following steps: S131b. Set the second threshold according to the evaluation requirements.

[0050] S132b. When the standard deviation of the request response time exceeds the second threshold, determine the performance evaluation test to which the key performance evaluation indicator belongs as an abnormal test.

[0051] Specifically, the standard deviation of the request response time is used to quantify the dispersion degree of the request response time in a single test, and the second threshold is dynamically set according to the evaluation requirements. Exemplarily, the evaluation requirements include at least one of the server stability requirements, business scenario types, or historical data. When setting the second threshold, combine the server hardware configuration and business scenario complexity, and determine the second threshold in combination with the distribution characteristics of the standard deviation of the request response time in historical data. When the standard deviation of the request response time exceeds the second threshold, it indicates that the fluctuation of the request response time exceeds the tolerance range of performance stability, triggering the determination of an abnormal test and determining the performance evaluation test as an abnormal test.

[0052] Exemplarily, Figure 11 Shows an example flowchart of the method for determining abnormal tests provided by an embodiment of the present application. Refer to Figure 11, at the beginning of the test, first determine the test conditions, namely the number of single-threads Cx and the concurrent number of single requests Ty, use the ramp-up test module to determine the minimum total concurrent number of single requests Q(Cx, Ty), and obtain the key performance evaluation indicators through the stress test module to determine abnormal tests. When the test time exceeds the first threshold or the standard deviation of the request response time exceeds the second threshold, it is determined as an abnormal test. The abnormal test rate and the non-abnormal throughput rate are counted, and the test ends after all performance evaluation tests are completed.

[0053] In the embodiment of the present application, by identifying the performance evaluation tests with large fluctuations in request response time, the performance evaluation tests with unstable request response time are determined as abnormal tests. After excluding the abnormal tests, the non-abnormal throughput rate is calculated, which further ensures the reliability and accuracy of performance evaluation.

[0054] Figure 12 shows a schematic flow chart of determining abnormal tests and non-abnormal throughput rates provided by an embodiment of the present application. Refer to Figure 12 , S10: Traverse each single-thread number in the thread number set and each single request concurrent number in the request concurrent number set in sequence, including the following steps: S11B: Obtain the number of server CPUs and the evaluation requirements.

[0055] S12B: Set a continuous integer interval for the request concurrent number set according to the evaluation requirements.

[0056] S13B: Take the number of server CPUs as an element value in the thread number set, and determine other elements in the thread number set according to the evaluation requirements to generate the thread number set.

[0057] S14B: Sample the continuous integer interval at a preset sampling interval within the continuous integer interval of the request concurrent number set to generate the request concurrent number set.

[0058] S15B: Traverse all combinations of the thread number set and the request concurrent number set in sequence, and use the thread number and the request concurrent number in each combination as the single-thread number and the single request concurrent number respectively to perform performance evaluation tests under the configuration of the single-thread number and the single request concurrent number.

[0059] Specifically, the number of server CPUs is set as the benchmark element of the thread number set to ensure that the number of threads matches the actual hardware resources. The evaluation requirements are used to dynamically adjust the range of request concurrency. Based on the number of server CPUs, the thread number set is further expanded according to the evaluation requirements. Exemplarily, the number of server CPUs can be the maximum thread value in the thread number set or an element value in the thread number set. The request concurrency set is determined by a continuous integer interval and a preset sampling interval. By constraining the thread number set and the request concurrency set within the range of the server's actual resources and evaluation requirements, it is possible to avoid the distortion of the performance evaluation test results caused by hyper-threading and balance the test efficiency and the comprehensiveness of test scenario coverage.

[0060] Figure 13 FIG. shows a flowchart of a method for drawing a stable critical curve provided by an embodiment of the present application. Refer to Figure 13 , S20. Perform a linear operation on the exception test rate and the non-exception throughput rate under each single-thread number and each single-request concurrency number to obtain the throughput stability loss rate. According to the throughput stability loss rate, construct a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate, including the following steps: S21. Determine the throughput stability loss rate according to the following formula:

[0061] where represents x request concurrency numbers, represents y thread numbers, represents the throughput stability loss rate under the configuration of x request concurrency numbers and y thread numbers, represents the non-exception throughput rate under the configuration of x request concurrency numbers and y thread numbers, represents the exception test rate under the configuration of x request concurrency numbers and y thread numbers.

[0062] S22. Draw a heat map of the throughput stability loss rate under each single-thread number and each single-request concurrency number configuration.

[0063] S23. Draw a stable critical curve according to a preset fitting segmentation algorithm and the heat map of the throughput stability loss rate under each single-thread number and each single-request concurrency number configuration.

[0064] Specifically, the throughput stability loss rate is determined by linearly weighting the normal throughput rate and the abnormal test rate, reflecting the total loss degree of performance and stability. The heat map is drawn by using color gradient mapping for different loss rate intervals to visualize the data distribution. The preset fitting and segmentation algorithm can use the gradient fitting algorithm, gradient descent algorithm, polynomial regression algorithm, machine learning non-linear fitting or segmentation algorithm in the prior art. The generation process of the preset fitting and segmentation algorithm will not be elaborated in this application.

[0065] In the embodiment of the present application, by quantifying the throughput stability loss rate and visualizing the stable critical curve, the performance boundary under different numbers of threads and different request concurrency numbers is intuitively displayed, avoiding the limitations of isolated indicators in traditional performance evaluation, providing more comprehensive and accurate performance evaluation results, determining the minimum thread configuration or the maximum request concurrency number according to the stable critical curve, providing a specific guiding direction for optimizing resource allocation, and helping to maximize the system performance on the premise of ensuring the server stability.

[0066] Figure 14 The flowchart of the method for determining the abnormal test rate and the normal throughput rate provided by an embodiment of the present application is shown. Refer to Figure 14 , S10. Obtain the abnormal test rate and the normal throughput rate under each single number of threads and each single request concurrency number, including the following steps: S11C. Determine the abnormal test rate according to the following formula:

[0067] S12C. Determine the normal throughput rate according to the following formula:

[0068] Wherein, represents the abnormal test rate under the configuration of x request concurrency numbers and y numbers of threads, represents the number of tests where the key performance evaluation index exceeds the preset threshold, represents the total number of tests, represents the normal throughput rate under the configuration of x request concurrency numbers and y numbers of threads, represents the throughput rate at the kth time under the configuration of x request concurrency numbers and y numbers of threads, represents the abnormal test indication function, is 1 indicating the test where the key performance evaluation index does not exceed the preset threshold at the kth time, is 0 indicating the test where the key performance evaluation index exceeds the preset threshold.

[0069] S13C. When the deviation between the throughput rate where a key performance evaluation indicator does not exceed the preset threshold and the throughput rate without anomalies is greater than the set threshold, or when the deviation between the throughput rate where the key performance evaluation indicator does not exceed the preset threshold and the throughput rate without anomalies does not have the characteristics of a normal distribution, adjust the preset threshold.

[0070] Specifically, the abnormal test rate is used to quantify the stability risk, and the throughput rate without anomalies ensures the reliability of the data by filtering abnormal test data. When the key performance evaluation indicator does not exceed the preset threshold, that is, for all non-abnormal tests, when the deviation between the throughput rate of one non-abnormal test and the throughput rate without anomalies is greater than the set threshold, or when the deviation between the throughput rates of all non-abnormal tests and the throughput rate without anomalies does not have the characteristics of a normal distribution, it is impossible to accurately distinguish abnormal tests through the preset quality, which affects the accuracy of the stable understanding curve. By triggering the dynamic adjustment mechanism of the preset threshold, the accuracy of determining abnormal tests is ensured.

[0071] In the embodiments of the present application, by adjusting the preset threshold to optimize the evaluation benchmark for abnormal tests, the adaptability of performance evaluation is further optimized, the accuracy of the abnormal test rate and the throughput rate without anomalies is ensured, and the evaluation results are made more objective and reliable.

[0072] Figure 15 The structural schematic diagram of a multi-threaded service concurrent request performance evaluation system provided by an embodiment of the present application is shown. Refer to Figure 15 , an embodiment of the present application provides a multi-threaded service concurrent request performance evaluation system, which is applied to a multi-threaded service concurrent request performance evaluation method, and includes: A system control module 510, configured to sequentially traverse each single-thread number in the thread number set and each single-request concurrency number in the request concurrency number set, and obtain the abnormal test rate and the throughput rate without anomalies under each single-thread number and each single-request concurrency number configuration. The abnormal test rate represents the proportion of the number of tests where the key performance evaluation indicator exceeds the preset threshold in the total number of tests, and the throughput rate without anomalies represents the statistical value of the throughput rate where the key performance evaluation indicator does not exceed the preset threshold.

[0073] A statistical estimation module 520, configured to perform a linear operation on the abnormal test rate and the throughput rate without anomalies under each single-thread number and each single-request concurrency number to obtain a throughput stability loss rate, and construct a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate according to the throughput stability loss rate. The stable critical curve represents the critical point of the balance relationship between the multi-threaded service concurrent request performance and stability under each single-thread number and each single-request concurrency number configuration.

[0074] In some embodiments, the system control module 510 is further configured to: Starting from the initial total request concurrency, it is increased step by step according to a preset step size to obtain the total request concurrency. At each total request concurrency, record the throughput rate configured for the single-thread count and the single-request concurrency. When the fluctuation range of the throughput rate is less than the fluctuation threshold, determine the minimum value among all the total request concurrencies corresponding to the maximum throughput rate, and generate the single total request concurrency.

[0075] Execute N performance evaluation tests according to the single-thread count, the single-request concurrency, and the single total request concurrency to obtain N test results, and each test result includes at least the throughput rate and the key performance evaluation indicators.

[0076] When the key performance evaluation indicators exceed the preset threshold, determine the performance evaluation test to which the key performance evaluation indicators belong as an abnormal test, calculate the proportion of the number of abnormal tests in the total number of N performance evaluation tests to obtain the abnormal test rate, and, determine the non-abnormal throughput rate according to the throughput rates of the non-abnormal tests in the N performance evaluation tests.

[0077] In some embodiments, the system control module 510 is further configured to: Obtain the number of server CPUs and the evaluation requirements.

[0078] According to the evaluation requirements, set a continuous integer interval for the request concurrency set.

[0079] Take the number of server CPUs as an element value in the thread count set, and determine other elements in the thread count set according to the evaluation requirements to generate the thread count set.

[0080] Sample the continuous integer interval at a preset sampling interval within the continuous integer interval of the request concurrency set to generate the request concurrency set.

[0081] Traverse all combinations of the thread count set and the request concurrency set in sequence, and respectively use the thread count and the request concurrency in each combination as the single-thread count and the single-request concurrency, and execute the performance evaluation test under the configuration of the single-thread count and the single-request concurrency.

[0082] In some embodiments, the statistical estimation module 520 is further configured to: Determine the throughput stability loss rate according to the following formula:

[0083] Wherein, represents x request concurrencies, represents y thread counts, represents the throughput stability loss rate under the configuration of x request concurrencies and y thread counts, represents the non-abnormal throughput rate under the configuration of x request concurrencies and y thread counts, It represents the exception test rate under the configuration of x request concurrency numbers and y thread numbers.

[0084] Draw a heat map of the throughput stability loss rate under each single thread number and each single request concurrency number configuration.

[0085] According to the preset fitting segmentation algorithm and the heat map of the throughput stability loss rate under each single thread number and each single request concurrency number configuration, draw a stable critical curve.

[0086] Figure 16 It shows a schematic structural diagram of a multi-threaded service concurrent request performance evaluation system provided by another embodiment of the present application. Refer to Figure 16 , the multi-threaded service concurrent request performance evaluation system further includes: The first determination module 530 is used to determine the corresponding single thread number on the stable critical curve as the minimum thread configuration when the preset target concurrency number is determined.

[0087] The second determination module 540 is used to determine the corresponding single request concurrency number on the stable critical curve as the maximum request concurrency number when the preset target thread number is determined.

[0088] In some embodiments, the system control module 510 is further used for: Determine the exception test rate according to the following formula:

[0089] Determine the no-exception throughput rate according to the following formula:

[0090] Wherein, represents the exception test rate under the configuration of x request concurrency numbers and y thread numbers, represents the number of tests where the key performance evaluation index exceeds the preset threshold, represents the total number of tests, represents the no-exception throughput rate under the configuration of x request concurrency numbers and y thread numbers, represents the throughput rate at the k-th time under the configuration of x request concurrency numbers and y thread numbers, represents the exception test indication function, being 1 indicates the test where the key performance evaluation index does not exceed the preset threshold at the k-th time, being 0 indicates the test where the key performance evaluation index exceeds the preset threshold.

[0091] In the case where the deviation between the throughput rate at which a key performance evaluation index does not exceed a preset threshold and the anomaly-free throughput rate is greater than a set threshold, or in the case where the deviation between the throughput rate at which a key performance evaluation index does not exceed a preset threshold and the anomaly-free throughput rate does not have a normal distribution characteristic, adjust the preset threshold.

[0092] In one embodiment, the key performance evaluation index at least includes a test time, and the preset threshold at least includes a first threshold.

[0093] The system control module 510 is further configured to: Set the first threshold according to the concurrent number of total requests per time.

[0094] In the case where the test time exceeds the first threshold, determine the performance evaluation test to which the key performance evaluation index belongs as an abnormal test.

[0095] In some embodiments, the key performance evaluation index at least includes a standard deviation value of a request response time, and the preset threshold at least includes a second threshold.

[0096] The system control module 510 is further configured to: Set the second threshold according to an evaluation requirement.

[0097] In the case where the standard deviation value of the request response time exceeds the second threshold, determine the performance evaluation test to which the key performance evaluation index belongs as an abnormal test.

[0098] For the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple. For related parts, refer to the partial description of the method embodiment.

[0099] Each embodiment in this specification is described in a progressive manner. Each embodiment focuses on the differences from other embodiments. For the same or similar parts among the embodiments, refer to each other.

[0100] Figure 17 The structural schematic diagram of an electronic device provided by an embodiment of the present application is shown. Refer to Figure 17 , an embodiment of the present application further provides an electronic device, including: A processor.

[0101] A memory for storing executable instructions of the processor.

[0102] Wherein, the processor is configured to execute instructions to implement a method for evaluating the performance of concurrent requests of a thread service.

[0103] In this embodiment, the computer device includes a processor, a memory, and a network interface connected through a system bus.

[0104] Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store data samples. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, it implements any one of the advanced anti-detection attack methods based on GPU memory loading.

[0105] Those skilled in the art can understand that Figure 17 the structure shown in

[0106] is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0107] The above computer-readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory, electrically erasable programmable read-only memory, erasable programmable read-only memory, programmable read-only memory, read-only memory, magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium accessible by a general-purpose or special-purpose computer.

[0108] Optionally, the readable storage medium is coupled to the processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application-specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0109] This application embodiment also provides a computer program product. The computer program product includes a computer program. When the computer program is executed by the processor, it implements any one of the advanced anti-detection attack methods based on GPU memory loading.

[0110] Those skilled in the art should understand that the embodiments of the present invention can be provided as a method, a system, or a computer program product. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present invention can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk memory, CD-ROM, optical memory, etc.) that contain computer-usable program code.

[0111] The present invention is described with reference to the flowcharts and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the present invention. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be realized by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices produce means for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0112] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing devices to work in a specific manner, such that the instructions stored in the computer-readable memory produce a manufactured article including instruction means that realizes the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0113] These computer program instructions can also be loaded onto a computer or other programmable data processing devices, such that a series of operation steps are executed on the computer or other programmable devices to generate a computer-implemented process, so that the instructions executed on the computer or other programmable devices provide steps for realizing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 or multiple blocks.

[0114] Although the preferred embodiments of the present invention have been described, those skilled in the art can make additional changes and modifications once they learn the basic creative concepts. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications that fall within the scope of the present invention.

[0115] Obviously, those skilled in the art can make various modifications and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

[0116] The above has introduced in detail a method, system, device and storage medium for evaluating the performance of concurrent requests for multi-threaded services provided by this application. Specific examples are used in this article to elaborate on the principle and implementation manner of this application. The description of the above embodiments is only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manner and application scope. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for evaluating the concurrent request performance of a multi-threaded service, characterized in that, Including the following steps: Traverse each single-thread number in the thread number set and each single-request concurrency number in the request concurrency number set in sequence, to obtain the exception test rate and the no-exception throughput rate under each configuration of the single-thread number and the single-request concurrency number. The exception test rate represents the proportion of the number of tests in which the key performance evaluation index exceeds the preset threshold in the total number of tests, and the no-exception throughput rate represents the statistical value of the throughput rate in which the key performance evaluation index does not exceed the preset threshold; Perform a linear operation on the exception test rate and the no-exception throughput rate under each single-thread number and each single-request concurrency number to obtain the throughput stability loss rate. According to the throughput stability loss rate, construct a stable critical curve with the single-request concurrency number as the abscissa and the single-thread number as the ordinate. The stable critical curve represents the critical point of the balance relationship between the multi-threaded service concurrent request performance and stability under each configuration of the single-thread number and the single-request concurrency number.

2. The method according to claim 1, wherein The step of obtaining the exception test rate and the no-exception throughput rate under each configuration of the single-thread number and the single-request concurrency number includes the following steps: Starting from the initial total request concurrency number, increase it step by step according to the preset step length to obtain the total request concurrency number. Under each total request concurrency number, record the throughput rate of the configuration of the single-thread number and the single-request concurrency number. When the fluctuation range of the throughput rate is less than the fluctuation threshold, determine the minimum value among all the total request concurrency numbers corresponding to the maximum throughput rate, and generate the single total request concurrency number; Execute N performance evaluation tests according to the single-thread number, the single-request concurrency number, and the single total request concurrency number to obtain N test results. Each test result includes at least the throughput rate and the key performance evaluation index; When the key performance evaluation index exceeds the preset threshold, determine the performance evaluation test to which the key performance evaluation index belongs as an exception test, calculate the proportion of the number of the exception tests in the total number of the N performance evaluation tests to obtain the exception test rate, and determine the no-exception throughput rate according to the throughput rate of the non-exception tests in the N performance evaluation tests.

3. The method according to claim 1, characterized in that, The step of traversing each single-thread number in the thread number set and each single-request concurrency number in the request concurrency number set in sequence includes the following steps: Obtain the number of server CPUs and the evaluation requirements; Set the continuous integer interval of the request concurrency number set according to the evaluation requirements; Take the number of server CPUs as an element value in the thread number set, and determine other elements in the thread number set according to the evaluation requirements to generate the thread number set; Sample the continuous integer interval in the continuous integer interval of the request concurrency number set at a preset sampling interval to generate the request concurrency number set; Traverse all combinations of the thread number set and the request concurrency number set in sequence, and respectively use the thread number and the request concurrency number in each combination as the single-thread number and the single-request concurrency number, and execute the performance evaluation test under the configuration of the single-thread number and the single-request concurrency number.

4. The method according to claim 1, wherein Perform a linear operation on the exception test rate and the exception-free throughput rate for each single-thread count and each single-request concurrency count to obtain the throughput stability loss rate. According to the throughput stability loss rate, construct a stable critical curve with the single-request concurrency count as the abscissa and the single-thread count as the ordinate, including the following steps: Determine the throughput stability loss rate according to the following formula: Among them, represents the number of concurrent requests of x, represents the number of threads of y, represents the throughput stability loss rate under the configuration of the number of concurrent requests of x and the number of threads of y, represents the throughput rate without exceptions under the configuration of the number of concurrent requests of x and the number of threads of y, represents the exception test rate under the configuration of the number of concurrent requests of x and the number of threads of y; Plot a heat map of the throughput stability loss rate for each configuration of the single-thread count and the single-request concurrency count; Draw the stable critical curve according to a preset fitting and segmentation algorithm and the heat map of the throughput stability loss rate for each configuration of the single-thread count and the single-request concurrency count; The method further includes: When a preset target concurrency count is determined, determine the corresponding single-thread count on the stable critical curve as the minimum thread configuration; When a preset target thread count is determined, determine the corresponding single-request concurrency count on the stable critical curve as the maximum request concurrency.

5. The method according to claim 1, characterized in that The obtaining of the exception test rate and the exception-free throughput rate for each configuration of the single-thread count and the single-request concurrency count includes the following steps: Determine the exception test rate according to the following formula: Determine the exception-free throughput rate according to the following formula: Among them, represents the exception test rate under the configuration of x request concurrency and y thread counts, represents the number of tests where the key performance evaluation indicators exceed the preset threshold, represents the total number of tests, represents the throughput rate without exceptions under the configuration of x request concurrency and y thread counts, represents the throughput rate at the k-th time under the configuration of x request concurrency and y thread counts, represents the exception test indication function, being 1 indicates the test where the key performance evaluation indicators at the k-th time do not exceed the preset threshold, being 0 indicates the test where the key performance evaluation indicators exceed the preset threshold; In the case where the deviation between the throughput rate at which a key performance evaluation indicator does not exceed a preset threshold and the exception-free throughput rate is greater than a set threshold, or in the case where the deviation between the throughput rate at which the key performance evaluation indicator does not exceed a preset threshold and the exception-free throughput rate does not have a normal distribution characteristic, adjust the preset threshold.

6. The method according to claim 2, wherein The key performance evaluation indicator at least includes the test time, and the preset threshold at least includes a first threshold; In the case where the key performance evaluation indicator exceeds a preset threshold, determining the performance evaluation test to which the key performance evaluation indicator belongs as an exception test includes the following steps: Set the first threshold according to the total single-request concurrency count; In the case where the test time exceeds the first threshold, determine the performance evaluation test to which the key performance evaluation indicator belongs as an exception test.

7. The method according to claim 2, wherein The key performance evaluation indicator at least includes the standard deviation of the request response time, and the preset threshold at least includes a second threshold; In the case where the key performance evaluation indicator exceeds a preset threshold, determining the performance evaluation test to which the key performance evaluation indicator belongs as an exception test includes the following steps: Set the second threshold according to the evaluation requirements; In the case where the standard deviation of the request response time exceeds the second threshold, determine the performance evaluation test to which the key performance evaluation indicator belongs as an exception test.

8. A multi-threaded service concurrent request performance evaluation system, characterized in that, Applied to the multi-threaded service concurrent request performance evaluation method according to any one of claims 1-7, the system includes: A system control module is used to sequentially traverse each single-thread number in the set of thread numbers and each single-request concurrency number in the set of request concurrency numbers, so as to obtain the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number configuration. The abnormal test rate represents the proportion of the number of tests in which the key performance evaluation index exceeds the preset threshold to the total number of tests, and the non-abnormal throughput rate represents the statistical value of the throughput rate in which the key performance evaluation index does not exceed the preset threshold. A statistical estimation module is used to perform a linear operation on the abnormal test rate and the non-abnormal throughput rate under each single-thread number and each single-request concurrency number to obtain the throughput stability loss rate. According to the throughput stability loss rate, a stable critical curve is constructed with the single-request concurrency number as the abscissa and the single-thread number as the ordinate. The stable critical curve represents the critical point of the balance relationship between the multi-threaded service concurrent request performance and stability under each single-thread number and each single-request concurrency number configuration.

9. An electronic device, characterized in that, It includes: A processor; A memory for storing executable instructions of the processor; Wherein, the processor is configured to execute the instructions to implement the thread service concurrent request performance evaluation method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, When the instructions in the computer-readable storage medium are executed by the processor of the terminal, the terminal can execute the thread service concurrent request performance evaluation method according to any one of claims 1 to 7.

Images