A method for testing server hardware performance

Abstract

The invention discloses a method for testing server hardware performance, and relates to the technical field of hardware performance testing. When screening software for final testing, key indicators such as CPU peak usage rate, memory usage peak, and software crash times corresponding to the software are obtained, and substituted into a calculation formula to obtain an evaluation value of the impact on hardware performance. This quantification method clearly numerically represents the degree of influence of fuzzy software on hardware, making the testing process more scientific. Through this method, it can be screened out for targeted testing, rather than general testing of all possible software, thereby improving the fit between performance testing and actual usage. During the testing process, not only the performance of the hardware in single-unit operation, such as the performance of each final test software running alone, but also the performance in combined operation, such as the performance of multiple final test software running in combination, is considered.

Description

Technical Field

[0001] The present invention relates to the technical field of hardware performance testing, and in particular to a method for testing server hardware performance. Background Art

[0002] As the target manufacturing enterprises continue to increase their digitalization, their business operations are increasingly dependent on the various services provided by servers. There are many types of servers in the enterprise, which run a large number of different types of software to support different departments and business processes. In the daily operation process, the performance of server hardware directly affects the continuity, efficiency and user experience of the business. Therefore, a server hardware performance testing method is needed.

[0003] Prior art, such as the invention patent application with announcement number CN114595132A, discloses a method and system for automatically detecting server hardware performance based on benchmark test indicators, including the following steps: first read the benchmark test indicator library, remotely and automatically obtain the target server hardware configuration, generate an indicator test estimate, create a server performance indicator detection script, remotely execute the detection script, structure the performance indicator detection results, and combine to generate a server performance detection report. The method and system for automatically detecting server hardware performance based on benchmark test indicators proposed in the present invention realizes automatic testing of server performance under the model of benchmark test indicators, obtains indicator data and analyzes and processes indicator data, automatically generates detection test scripts in the process, collects server information in combination with the application server BMCWebAPI interface, and automatically generates test reports, effectively meeting the needs of server hardware performance benchmark indicator testing and realizing rapid automation of the test process.

[0004] With respect to the above scheme, there are the following technical problems: 1. The existing technology lacks a screening method, and the test may perform performance tests on all software installed on the server. This will cause the test scope to be too broad, including many software that have little impact on the performance of the server hardware or are rarely used in actual business scenarios. For example, some users occasionally install small tool software for personal testing, which will also be included in the test scope, resulting in a large amount of irrelevant data generated during the test process, increasing the time and workload of the test, reducing the test efficiency, and may not be able to focus on the core software that is frequently used in the actual production environment of the enterprise, resulting in the test results not accurately reflecting the hardware performance of the server in the real business scenario. For example, the core business of the enterprise is based on a specific database management software and related middleware, but the test does not focus on the impact of these software on the hardware, but instead tests various non-critical software. The hardware performance evaluation results obtained in this way will have a large deviation from the performance in the actual business operation.

[0005] Existing technologies may only focus on the performance of hardware when a single software is running, ignoring the complex situation of multiple software combinations running in the actual operation of the server. For example, in an enterprise server environment, Web server software and database software usually work at the same time, and the data interaction and resource sharing between them will have a mutual impact on hardware performance. If you only test the performance of the hardware when running Web server software or database software alone, you will not be able to find hardware bottlenecks that may occur when the software combination is running, such as memory contention, CPU scheduling conflicts, and other problems.

[0006] The existing technology does not have a method to quantify the impact of software on hardware performance, so it is difficult to determine the role of different software in the process of server performance degradation. For example, when a server has a performance problem, it is impossible to determine whether it is due to a high-load computing software that causes CPU overload, or a disk I / O bottleneck caused by frequently read and written storage software. This makes it lack of targeted direction when optimizing server performance or solving problems. There is no software screening mechanism based on quantitative indicators. When selecting software for testing, it may be blind, and it is impossible to accurately select the software that has the greatest impact on hardware performance for key testing. This may cause testing resources to be wasted on software that has less impact on hardware performance, while ignoring the key software that may actually cause performance problems. Summary of the invention

[0007] In view of the above-mentioned technical deficiencies, an object of the present invention is to provide a method for testing server hardware performance.

[0008] In order to solve the above technical problems, the present invention adopts the following technical solutions: The present invention provides a method for testing server hardware performance, including: Step 1, obtaining type usage software: obtaining the usage software corresponding to each user in each type of server in the historical period of the target production enterprise, so as to perform initial test screening on the usage software corresponding to each user in each type of server, and then screen out each type of initial test usage software corresponding to each type of server.

[0009] Step 2: Final test screening: Based on the software used in the initial tests of each type of server, the software used in the initial tests of each type of server is screened to obtain the software used in the final tests of each type of server.

[0010] Step 3. Acquisition of hardware performance test data: Input the final test software corresponding to each type of server into the hardware performance test platform of each type of server, so as to perform performance test on each type of hardware in each type of server in the current cycle of the target production enterprise, and set several collection time points during the test process, and then analyze the single performance evaluation value corresponding to the single operation of each type of final test software and the combined performance evaluation value corresponding to the combined operation of each type of final test software in each type of server at each collection time point.

[0011] Step 4, judgment of hardware performance: according to the single performance evaluation value corresponding to the single operation of each type of hardware in each type of server using the software for the final test of each type at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of software for the final test of each type, analyze the performance evaluation value corresponding to each hardware in each type of server, and then judge whether the performance of each hardware in each type of server is qualified.

[0012] Preferably, the software used corresponding to each user in each type of server is initially tested and screened, and the specific screening process is as follows: A1. Collect the software used corresponding to each user in each type of server within the historical period from the server management system of the target production enterprise, classify the collected software used corresponding to each user in each type of server, obtain the software used corresponding to each user in each type of server, and then count the usage frequency of each type of software used corresponding to each type of server.

[0013] A2. Compare the usage frequency of each type of software corresponding to each type of server with the usage frequency of the set initial test software. If the usage frequency of a certain type of software corresponding to a certain type of server is greater than or equal to the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is representative of the test. If the usage frequency of a certain type of software corresponding to a certain type of server is less than the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is not representative of the test, and the types of software corresponding to the types of servers that are representative of the test are recorded as the initial test software of each type.

[0014] Preferably, the software used for the initial test corresponding to each type of server is finally screened, and the specific screening process is as follows: B1. Obtain the hardware performance evaluation value corresponding to each type of software used for the initial test in each type of server, and arrange the hardware performance evaluation value corresponding to each type of software used for the initial test in each type of server in order from large to small.

[0015] B2. The software used in the initial test of the ten types of servers that have the highest impact on the hardware performance evaluation value are recorded as the final test software of each type.

[0016] Preferably, the hardware performance evaluation value corresponding to each type of software used in the initial test of each type of server is obtained, and the specific acquisition process is as follows: the CPU peak usage rate, memory usage peak value and software crash number corresponding to each type of software used in the initial test of each type of server are obtained, and the CPU peak usage rate, memory usage peak value and software crash number corresponding to each type of software used in the initial test of each type of server are recorded as , and , q represents the number corresponding to each type of server, q is a positive integer, w represents the number corresponding to the software used for the initial test of each type, w is a positive integer, substitute into the calculation formula:

[0017] In the above example, we can get the hardware performance evaluation value corresponding to the initial test software used in each type of server. ,in, , , They are the standard CPU peak usage, standard memory usage peak, and standard software crash times corresponding to the set software. , , They are the weight factors corresponding to the peak CPU usage rate of the set software, the weight factors corresponding to the peak memory usage rate, and the weight factors corresponding to the number of software crashes. , , They are the adjustment factors corresponding to the peak CPU usage of the set software, the adjustment factors corresponding to the peak memory usage, and the adjustment factors corresponding to the number of software crashes. Represents a natural constant.

[0018] Preferably, the final test software corresponding to each type of server is input into the hardware performance test platform of each type of server for performance testing. The specific test process is as follows: C1. Install each hardware of each type of server in the current period of the target production enterprise into the corresponding test platform, and install the corresponding final test software of each type on the test platform of each type of server.

[0019] C2. Perform a single-run test on each type of hardware in each type of server first. That is, each time the test platform runs each type of server, only one type of final test software is run on the test server. Then, at each collection time point, the server hardware performance monitoring tool is used to collect the single-run performance data of each type of hardware in each type of server when the software is run separately. The single-run performance data includes single-run response time, single-run power consumption and single-run heat dissipation efficiency.

[0020] C3. After each hardware in each type of server completes the individual operation test, a combined operation test is performed on each hardware in each type of server. The combined operation test is performed according to the combined operation test of each type of final test software that appears in the final test software. That is, each time the test platform runs each type of server, only one type of final test software combination is run on the test server. Then, at each collection time point, the server hardware performance monitoring tool is used to collect the combined performance data corresponding to the combined operation of each type of hardware in each type of server using each type of final test software. The combined performance data includes combined response time, combined power consumption and combined heat dissipation efficiency.

[0021] Preferably, the analysis at each collection time point of the single performance evaluation value corresponding to the single operation of each type of final test software used in each type of hardware in each type of server is performed, and the specific analysis process is as follows: the single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of final test software used in each type of hardware in each type of server at each collection time point are obtained, and the single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of final test software used in each type of hardware in each type of server at each collection time point are input into the single performance evaluation analysis model, and the single performance evaluation value corresponding to the single operation of each type of final test software used in each type of hardware in each type of server at each collection time point is output.

[0022] Preferably, the combined performance evaluation value corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server is analyzed at each collection time point, and the specific analysis process is as follows: the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point are obtained, and the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point are input into the combined performance evaluation analysis model, and the combined performance evaluation value corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point is output.

[0023] Preferably, the specific analysis process of judging whether the performance of each hardware in each type of server is qualified is as follows: inputting the single performance evaluation value corresponding to the single operation of each type of final test software and the combined performance evaluation value corresponding to the combined operation of each type of final test software in each type of server at each collection time point into the performance evaluation value evaluation model, and outputting the performance evaluation value results corresponding to each hardware in each type of server;

[0024] The performance evaluation value results include values ​​of 1 and -1. When the performance evaluation value result corresponding to a certain hardware in a certain type of server is 1, it indicates that the performance of the hardware in this type of server is qualified. Conversely, when the performance evaluation value result corresponding to a certain hardware in a certain type of server is -1, it indicates that the performance of the hardware in this type of server is unqualified. If the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback will be given.

[0025] Preferably, the expression of the performance evaluation value evaluation model is:

[0026] , where It represents the performance evaluation result corresponding to the r-th hardware in the q-th type server. represents the performance evaluation value corresponding to the rth hardware in the qth type server, is the set performance evaluation value threshold, q represents the number corresponding to each type of server, q is a positive integer, r represents the number corresponding to each hardware, r is a positive integer.

[0027] The single performance evaluation value corresponding to the single operation of each hardware in each type of server at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of final test software are recorded as and , substitute into the calculation formula:

[0028] In the above example, we can get the performance evaluation value of each hardware in each type of server. , where i represents the number corresponding to each acquisition time point, i is a positive integer, f represents the number corresponding to each type of final test using a single software run, f is a positive integer, h represents the number corresponding to each type of final test using a combination of software runs, h is a positive integer, , They are the standard single performance evaluation value corresponding to the final test of the specified hardware using the software single operation, and the standard combination performance evaluation value corresponding to the final test using the software combination operation. , They are respectively the weight factor corresponding to the single performance evaluation value of the set hardware when the software is run individually in the final test, and the weight factor corresponding to the combined performance evaluation value when the software is run in combination in the final test.

[0029] Preferably, if the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback is performed, and the specific early warning process is as follows: when the performance evaluation result of a certain hardware in a certain type of server is monitored to be unqualified, the system will immediately activate the early warning mechanism. At this time, an early warning message will be sent to relevant personnel responsible for server operation and maintenance. The early warning message includes the server type and hardware name.

[0030] The beneficial effects of the present invention are as follows: 1. In the embodiment of the present invention, by collecting, classifying and screening the software used by users in various types of servers in the historical period of the target production enterprise, the software used for the initial test is first determined, and then the software used for the final test is further screened out. This method makes the test software closely related to the actual use of the server, avoids the test of irrelevant software, and can more accurately reflect the hardware performance of the server in the real business scenario. For example, in an enterprise, if a database server mainly runs a specific version of database management software, it can be screened out for targeted testing through this method, rather than general testing of all possible software, thereby improving the fit between performance testing and actual use. During the test process, not only the performance of the hardware in single operation, such as the performance of each final test software running alone, but also the performance in combined operation, such as the performance of multiple final test software running in combination. This can comprehensively evaluate the performance of the hardware in complex business scenarios, because in the actual server operating environment, multiple software often run simultaneously and affect each other. For example, on a server that is simultaneously running Web server software, database software, and middleware, this comprehensive evaluation can accurately detect the performance bottlenecks of the hardware when processing the collaborative work of multiple software, thereby improving the accuracy of hardware performance evaluation.

[0031] 2. In the embodiment of the present invention, when screening the software for the final test, the key indicators such as the CPU peak usage rate, the memory usage peak and the number of software crashes corresponding to the software are obtained, and the evaluation value of the impact on the hardware performance is obtained by substituting them into the calculation formula. This quantification method clearly expresses the degree of influence of the fuzzy software on the hardware, making the test process more scientific. For example, by setting the weight factor and the adjustment factor, the weight of the number of software crashes can be increased according to the degree of attention paid by the enterprise to different factors, such as high requirements for server stability, and the impact of each software on the hardware performance can be accurately measured, providing a reliable basis for subsequent testing and evaluation. The performance data of the collected hardware under different operating conditions, such as the single response time, single power consumption, single heat dissipation efficiency, combined response time, combined power consumption, combined heat dissipation efficiency, etc., are converted into quantifiable evaluation values ​​by using the single performance evaluation analysis model and the performance evaluation value evaluation model. By comparing with the set threshold, it is possible to accurately judge whether the hardware performance is qualified. This model-based method standardizes and objectifies the evaluation process, reduces the subjectivity and error of human judgment, and provides a scientific basis for the performance judgment of server hardware.

[0032] 3. According to the embodiment of the present invention, the actual performance of the server hardware under different business loads can be understood through accurate hardware performance testing. Enterprises can reasonably configure server resources based on this information to avoid over-configuration or under-configuration of hardware resources. For example, if it is found that the hardware of a server still has a large margin under the current business load, it can be considered to migrate the business of some other servers to improve the utilization rate of hardware resources; conversely, if the hardware performance is close to the limit, the hardware upgrade can be planned in advance to ensure the smooth operation of the business. At the same time, the complete process from software acquisition, screening to hardware performance testing, evaluation and early warning is clearly defined, and each step has a detailed operation method and basis. This makes the testing process repeatable, and different testers can get similar results according to the same method. For example, newly employed operation and maintenance personnel can accurately test the performance of server hardware based on this method without relying on personal experience, ensuring the stability and reliability of the test quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0034] Figure 1 The present invention is a flowchart of the steps for implementing the method. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] Embodiments of the present invention Figure 1 As shown, a method for testing server hardware performance, step one, obtaining type usage software: obtaining each usage software corresponding to each user in each type of server in the historical period of the target production enterprise, and then performing initial test screening on each usage software corresponding to each user in each type of server, and then screening to obtain each type of initial test usage software corresponding to each type of server.

[0037] In a specific embodiment, the software used corresponding to each user in each type of server is initially tested and screened, and the specific screening process is as follows: A1. Collect the software used corresponding to each user in each type of server in the historical period from the server management system of the target production enterprise, classify the collected software used corresponding to each user in each type of server, obtain the software used corresponding to each user in each type of server, and then count the usage frequency of each type of software used corresponding to each type of server.

[0038] It should be noted that the types of software used include: testing operating systems, testing driver versions and applications, extracting software data used by various types of server users from the server management system of the target production enterprise, including information such as operating systems, driver versions and applications, cleaning the collected data, removing duplicate, incomplete or irrelevant information, ensuring the accuracy and consistency of the data, and dividing it into the following categories according to the type of software: Operating system: Identify and classify different operating systems, such as Windows, Linux, Unix, etc. Driver: Identify and classify various driver programs, such as network drivers, graphics card drivers, etc., and record their version information. Application: Identify and classify various applications, such as office software, databases, development tools, etc.

[0039] A2. Compare the usage frequency of each type of software corresponding to each type of server with the usage frequency of the set initial test software. If the usage frequency of a certain type of software corresponding to a certain type of server is greater than or equal to the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is representative of the test. If the usage frequency of a certain type of software corresponding to a certain type of server is less than the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is not representative of the test, and the types of software corresponding to the types of servers that are representative of the test are recorded as the initial test software of each type.

[0040] Step 2: Final test screening: Based on the software used in the initial tests of each type of server, the software used in the initial tests of each type of server is screened to obtain the software used in the final tests of each type of server.

[0041] In a specific embodiment, the software used for the initial test corresponding to each type of server is finally screened, and the specific screening process is as follows: B1. Obtain the hardware performance evaluation value corresponding to the software used for the initial test in each type of server, and arrange the hardware performance evaluation value corresponding to the software used for the initial test in each type of server in order from large to small.

[0042] B2. The software used in the initial test of the ten types of servers that have the highest impact on the hardware performance evaluation value are recorded as the final test software of each type.

[0043] In another specific embodiment, the hardware performance evaluation value corresponding to each type of software used in the initial test of each type of server is obtained, and the specific acquisition process is as follows: the CPU peak usage rate, memory usage peak value and software crash number corresponding to each type of software used in the initial test of each type of server are obtained, and the CPU peak usage rate, memory usage peak value and software crash number corresponding to each type of software used in the initial test of each type of server are recorded as , and , q represents the number corresponding to each type of server, q is a positive integer, w represents the number corresponding to the software used for the initial test of each type, w is a positive integer, substitute into the calculation formula:

[0044] In the above example, we can get the hardware performance evaluation value corresponding to the initial test software used in each type of server. ,in, , , They are the standard CPU peak usage, standard memory usage peak, and standard software crash times corresponding to the set software. , , They are the weight factors corresponding to the peak CPU usage rate of the set software, the weight factors corresponding to the peak memory usage rate, and the weight factors corresponding to the number of software crashes. , , They are the adjustment factors corresponding to the peak CPU usage of the set software, the adjustment factors corresponding to the peak memory usage, and the adjustment factors corresponding to the number of software crashes. Represents a natural constant.

[0045] It should be noted that , , are both greater than 0 and less than 1, , , Both are greater than 0 and less than 1.

[0046] It should also be noted that through the summary of a large amount of research data and experimental data. Professional organizations and research institutions set the standard CPU peak usage rate, standard memory usage peak, and standard software crash times for the software. At the same time, based on the professional knowledge and research basis of experts in the field, they discuss and confirm with industry organizations or professional institutions. Experts set the weight factors corresponding to the CPU peak usage rate of the software, the weight factors corresponding to the memory usage peak, and the weight factors corresponding to the number of software crashes based on their own experience and knowledge, and set the adjustment factors corresponding to the CPU peak usage rate of the software, the adjustment factors corresponding to the memory usage peak, and the adjustment factors corresponding to the number of software crashes.

[0047] In the embodiment of the present invention, by collecting, classifying and screening the software used by users in various types of servers in the historical period of the target production enterprise, the software used for the initial test is first determined, and then the software used for the final test is further screened out. This method makes the test software closely related to the actual use of the server, avoids the test of irrelevant software, and can more accurately reflect the hardware performance of the server in the real business scenario. For example, in an enterprise, if a database server mainly runs a specific version of database management software, it can be screened out by this method for targeted testing, rather than general testing of all possible software, thereby improving the fit between the performance test and the actual use situation. During the test process, not only the performance of the hardware in a single operation, such as each final test software running alone, but also the performance in a combined operation, such as a combination of multiple final test software running. This can comprehensively evaluate the performance of the hardware in complex business scenarios, because in the actual server operating environment, multiple software often run simultaneously and affect each other. For example, on a server that runs Web server software, database software and middleware at the same time, this comprehensive evaluation can accurately detect the performance bottleneck of the hardware when processing multiple software collaborative work, and improve the accuracy of hardware performance evaluation.

[0048] Step 3. Acquisition of hardware performance test data: Input the final test software corresponding to each type of server into the hardware performance test platform of each type of server, so as to perform performance test on each type of hardware in each type of server in the current cycle of the target production enterprise, and set several collection time points during the test process, and then analyze the single performance evaluation value corresponding to the single operation of each type of final test software and the combined performance evaluation value corresponding to the combined operation of each type of final test software in each type of server at each collection time point.

[0049] It should be noted that the final test software single run refers to the way in which only one type of final test software is run on the test server each time during the server hardware performance test. This running mode is mainly used to isolate the impact of different software on the hardware and evaluate the performance of each software on the server hardware separately.

[0050] It should also be noted that the final test uses software combination operation, which means that during the server hardware performance test, multiple software are run simultaneously according to the combination of each type of final test software in the actual business scenario. This operation mode is used to simulate the working state of the server in the real business environment and evaluate the comprehensive performance of the hardware when dealing with multiple software working together.

[0051] In a specific embodiment, the final test software corresponding to each type of server is input into the hardware performance test platform of each type of server for performance testing. The specific test process is as follows: C1. Install each hardware of each type of server in the current period of the target production enterprise into the corresponding test platform, and install the corresponding final test software of each type on the test platform of each type of server.

[0052] C2. Perform a single-run test on each type of hardware in each type of server first. That is, each time the test platform runs each type of server, only one type of final test software is run on the test server. Then, at each collection time point, the server hardware performance monitoring tool is used to collect the single-run performance data of each type of hardware in each type of server when the software is run separately. The single-run performance data includes single-run response time, single-run power consumption and single-run heat dissipation efficiency.

[0053] C3. After each hardware in each type of server completes the individual operation test, a combined operation test is performed on each hardware in each type of server. The combined operation test is performed according to the combined operation test of each type of final test software that appears in the final test software. That is, each time the test platform runs each type of server, only one type of final test software combination is run on the test server. Then, at each collection time point, the server hardware performance monitoring tool is used to collect the combined performance data corresponding to the combined operation of each type of hardware in each type of server using each type of final test software. The combined performance data includes combined response time, combined power consumption and combined heat dissipation efficiency.

[0054] It should be noted that appropriate monitoring points are set at the software code and hardware interface levels. For software, a timestamp is inserted at the code position where the task request is issued, and a timestamp is also inserted at the code position where the hardware completes the task and returns the result. These timestamps can be implemented through the logging function of the software or a dedicated performance monitoring tool to obtain the single response time and the combined response time. The hardware power consumption data is obtained by connecting to the power meter on the server power line or the power consumption monitoring chip on the server motherboard to obtain the single power consumption and the combined power consumption. The monitoring tool is used to record the start time of all software running simultaneously to issue task requests, and the end time of all tasks completing and returning the result. The difference between the two is the combined response time. Considering the task association, more accurate data can be obtained by simulating different load combinations and measuring multiple times. The total power consumption can be directly measured by the power monitoring function of the server power distribution unit or an external power meter. The software tool can also be used to read the estimated power consumption of each hardware component and add them together to obtain the total power consumption. Analyze the power consumption changes under different software combinations and loads, first integrate the data of each hardware temperature sensor, and calculate the average temperature rise value of the hardware. Determine the total energy consumption of the cooling system, considering the internal heat transfer and environmental factors of the chassis. According to the law of conservation of energy, the ratio of the heat taken away by the cooling system to the total heat generated by the hardware is calculated, and then the single heat dissipation efficiency and the combined heat dissipation efficiency are obtained.

[0055] In another specific embodiment, the individual performance evaluation values ​​corresponding to the individual operation of each type of final test software used in each type of hardware in each type of server are analyzed at each collection time point. The specific analysis process is as follows: the individual response time, individual power consumption and individual heat dissipation efficiency corresponding to the individual operation of each type of final test software used in each type of server at each collection time point are obtained, and the individual response time, individual power consumption and individual heat dissipation efficiency corresponding to the individual operation of each type of final test software used in each type of server at each collection time point are input into the individual performance evaluation analysis model, and the individual performance evaluation values ​​corresponding to the individual operation of each type of final test software used in each type of server at each collection time point are output.

[0056] It should be noted that the analysis process of the single performance evaluation value corresponding to the single operation of each type of hardware in each type of server in each type of final test software at each collection time point is as follows: the single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of hardware in each type of server in each type of final test software at each collection time point are normalized, and the single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of hardware in each type of server in each type of final test software at each collection time point after processing are recorded as , and , substitute into the analysis formula:

[0057] , get the single performance evaluation value corresponding to each hardware in each type of server at each collection time point in each type of final test using the software single operation , , , They are the weight coefficients corresponding to the single response time of the single software running in the set final test, the weight coefficients corresponding to the single power consumption, and the weight coefficients corresponding to the single heat dissipation efficiency. Among them, i represents the number corresponding to each collection time point, i is a positive integer, f represents the number corresponding to the single software running in the final test of each type, f is a positive integer, q represents the number corresponding to each type of server, q is a positive integer, and r represents the number corresponding to each hardware, r is a positive integer.

[0058] It should be noted that , , Both are greater than 0 and less than 1.

[0059] It should also be noted that, based on the expertise and research of experts in the field, and after discussion and confirmation with industry organizations or professional institutions, the experts set the weight coefficients corresponding to the single-unit response time, single-unit power consumption, and single-unit heat dissipation efficiency of the final test software based on their own experience and knowledge.

[0060] In a specific embodiment, the combined performance evaluation value corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server is analyzed at each collection time point. The specific analysis process is as follows: the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point are obtained, and the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point are input into the combined performance evaluation analysis model, and the combined performance evaluation value corresponding to the combined operation of each type of software used in each type of final test in each type of hardware in each type of server at each collection time point is output.

[0061] It should be noted that the analysis process of the combined performance evaluation value corresponding to the single running of each type of software used in each type of final test in each type of server at each collection time point is as follows: the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the single running of each type of software used in each type of final test in each type of server at each collection time point are normalized, and the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the single running of each type of software used in each type of final test in each type of server at each collection time point after processing are recorded as , and , substitute into the analysis formula:

[0062] , and obtain the combined performance evaluation values ​​corresponding to the software monomer operation of each type of hardware in each type of server at each collection time point , , , They are the weight coefficients corresponding to the response time of the combined operation of the software monomer for the set final test, the weight coefficients corresponding to the combined power consumption, and the weight coefficients corresponding to the combined heat dissipation efficiency, respectively. Among them, i represents the number corresponding to each collection time point, i is a positive integer, h represents the number corresponding to each type of software combination for the final test, h is a positive integer, q represents the number corresponding to each type of server, q is a positive integer, and r represents the number corresponding to each hardware, r is a positive integer.

[0063] It should be noted that , , Both are greater than 0 and less than 1.

[0064] It should also be noted that, based on the expertise and research of experts in the field, and after discussion and confirmation with industry organizations or professional institutions, the experts set the weight coefficients corresponding to the combined response time, combined power consumption, and combined heat dissipation efficiency of the software monomer operation combination according to their own experience and knowledge.

[0065] Step 4, judgment of hardware performance: according to the single performance evaluation value corresponding to the single operation of each type of hardware in each type of server using the software for the final test of each type at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of software for the final test of each type, analyze the performance evaluation value corresponding to each hardware in each type of server, and then judge whether the performance of each hardware in each type of server is qualified.

[0066] In a specific embodiment, the determination of whether the performance of each hardware in each type of server is qualified is performed by the following specific analysis process: the individual performance evaluation values ​​corresponding to the individual operation of each type of final test software used in each type of server at each collection time point and the combined performance evaluation values ​​corresponding to the combined operation of each type of final test software used are input into the performance evaluation value evaluation model, and the performance evaluation value results corresponding to each hardware in each type of server are output.

[0067] The performance evaluation value results include values ​​of 1 and -1. When the performance evaluation value result corresponding to a certain hardware in a certain type of server is 1, it indicates that the performance of the hardware in this type of server is qualified. Conversely, when the performance evaluation value result corresponding to a certain hardware in a certain type of server is -1, it indicates that the performance of the hardware in this type of server is unqualified. If the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback will be given.

[0068] In a specific embodiment, the expression of the performance evaluation value evaluation model is:

[0069] , where It represents the performance evaluation result corresponding to the r-th hardware in the q-th type server. represents the performance evaluation value corresponding to the rth hardware in the qth type server, is the set performance evaluation value threshold, q represents the number corresponding to each type of server, q is a positive integer, r represents the number corresponding to each hardware, r is a positive integer.

[0070] The single performance evaluation value corresponding to the single operation of each hardware in each type of server at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of final test software are recorded as and , substitute into the calculation formula:

[0071] In the above example, we can get the performance evaluation value of each hardware in each type of server. , where i represents the number corresponding to each acquisition time point, i is a positive integer, f represents the number corresponding to each type of final test using a single software run, f is a positive integer, h represents the number corresponding to each type of final test using a combination of software runs, h is a positive integer, , They are the standard single performance evaluation value corresponding to the final test of the specified hardware using the software single operation, and the standard combination performance evaluation value corresponding to the final test using the software combination operation. , They are respectively the weight factor corresponding to the single performance evaluation value of the set hardware when the software is run individually in the final test, and the weight factor corresponding to the combined performance evaluation value when the software is run in combination in the final test.

[0072] It should be noted that , Both are greater than 0 and less than 1.

[0073] It should also be noted that through the summary of a large amount of research data and experimental data, professional institutions and research institutions set the standard single performance evaluation value corresponding to the final test of the hardware using the single software running, and the standard combination performance evaluation value corresponding to the final test using the software combination running. At the same time, based on the professional knowledge and research basis of experts in the field, and discussed and confirmed with industry organizations or professional institutions, experts set the weight factors corresponding to the single performance evaluation value when the hardware is running the single software in the final test, and the weight factors corresponding to the combination performance evaluation value when the software is running in the final test.

[0074] In the embodiment of the present invention, when screening the software for the final test, the key indicators such as the CPU peak usage rate, the memory usage peak and the number of software crashes corresponding to the software are obtained, and the evaluation value of the impact on the hardware performance is obtained by substituting them into the calculation formula. This quantification method clearly expresses the degree of influence of the fuzzy software on the hardware, making the test process more scientific. For example, by setting the weight factor and the adjustment factor, the weight of the number of software crashes can be increased according to the importance of the enterprise to different factors, such as high requirements for server stability, and the impact of each software on the hardware performance can be accurately measured, providing a reliable basis for subsequent testing and evaluation. The performance data of the collected hardware under different operating conditions, such as the single response time, single power consumption, single heat dissipation efficiency, combined response time, combined power consumption, combined heat dissipation efficiency, etc., are converted into quantifiable evaluation values ​​by using the single performance evaluation analysis model and the performance evaluation value evaluation model. By comparing with the set threshold, it is possible to accurately judge whether the hardware performance is qualified. This model-based method standardizes and objectifies the evaluation process, reduces the subjectivity and error of human judgment, and provides a scientific basis for the performance judgment of server hardware.

[0075] In a specific embodiment, if the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback is performed. The specific early warning process is as follows: when the performance evaluation result of a certain hardware in a certain type of server is monitored to be unqualified, the system will immediately activate the early warning mechanism. At this time, an early warning message will be sent to the relevant personnel responsible for the server operation and maintenance. The early warning message includes the server type and hardware name.

[0076] According to the embodiment of the present invention, the actual performance of the server hardware under different business loads can be understood through accurate hardware performance testing. Enterprises can reasonably configure server resources based on this information to avoid over-configuration or under-configuration of hardware resources. For example, if it is found that the hardware of a server still has a large margin under the current business load, it can be considered to migrate the business of some other servers to improve the utilization rate of hardware resources; conversely, if the hardware performance is close to the limit, the hardware upgrade can be planned in advance to ensure the smooth operation of the business. At the same time, the complete process from software acquisition, screening to hardware performance testing, evaluation and early warning is clearly defined, and each step has a detailed operation method and basis. This makes the testing process repeatable, and different testers can get similar results according to the same method. For example, newly employed operation and maintenance personnel can accurately test the performance of server hardware based on this method without relying on personal experience, thereby ensuring the stability and reliability of the test quality.

[0077] The above contents are merely examples and explanations of the concept of the present invention. Those skilled in the art may make various modifications or additions to the specific embodiments described or replace them in a similar manner. As long as they do not deviate from the concept of the invention or exceed the scope defined in this specification, they shall all fall within the protection scope of the present invention.

Claims

1. A method for testing server hardware performance, characterized in that: include: Step 1: Obtaining software used by each type: Obtaining software used by each user in each type of server in the historical period of the target production enterprise, thereby performing initial testing and screening on the software used by each user in each type of server, and then screening out each type of initially tested software used by each type corresponding to each type of server; Step 2, final test screening: according to the software used in the initial test of each type of server corresponding to each type of server, the software used in the initial test of each type of server corresponding to each type of server is finally screened, and then the software used in the final test of each type of server corresponding to each type of server is screened; The initial tests of each type of server corresponding to each type are finally screened using software. The specific screening process is as follows: B1. Obtain the hardware performance impact evaluation values ​​corresponding to the software used in the initial test of each type of server, and arrange the hardware performance impact evaluation values ​​corresponding to the software used in the initial test of each type of server in order from large to small; B2. The ten types of software used in the initial test that have the highest impact on the hardware performance evaluation value of each type of server are recorded as the final test software used for each type; The specific acquisition process of obtaining the hardware performance evaluation value corresponding to the software used in the initial test of each type of server is as follows: Get the CPU peak usage rate, memory usage peak value, and software crash number corresponding to each type of software used in the initial test of each type of server, and record the CPU peak usage rate, memory usage peak value, and software crash number corresponding to each type of software used in the initial test of each type of server as , and , q represents the number corresponding to each type of server, q is a positive integer, w represents the number corresponding to the software used for the initial test of each type, w is a positive integer, substitute into the calculation formula: In the above example, we can get the hardware performance evaluation value corresponding to the initial test software used in each type of server. ,in, , , They are the standard CPU peak usage, standard memory usage peak, and standard software crash times corresponding to the set software. , , They are the weight factors corresponding to the peak CPU usage rate of the set software, the weight factors corresponding to the peak memory usage rate, and the weight factors corresponding to the number of software crashes. , , They are the adjustment factors corresponding to the peak CPU usage of the set software, the adjustment factors corresponding to the peak memory usage, and the adjustment factors corresponding to the number of software crashes. represents a natural constant; Step 3, acquisition of hardware performance test data: input each type of final test software corresponding to each type of server into each type of server hardware performance test platform, so as to perform performance test on each type of hardware in each type of server in the current cycle of the target production enterprise, and set several collection time points during the test process, and then analyze the single performance evaluation value corresponding to the single operation of each type of final test software and the combined performance evaluation value corresponding to the combined operation of each type of final test software in each type of server at each collection time point; Step 4, judgment of hardware performance: according to the single performance evaluation value corresponding to the single operation of each type of hardware in each type of server using the software for the final test of each type at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of software for the final test of each type, analyze the performance evaluation value corresponding to each hardware in each type of server, and then judge whether the performance of each hardware in each type of server is qualified.

2. A method for testing server hardware performance as claimed in claim 1, characterized in that: The initial test screening of each software used by each user in each type of server is as follows: A1. Collecting various software used by various users in various types of servers in the historical period from the server management system of the target manufacturing enterprise, classifying the collected various software used by various users in various types of servers, obtaining various types of software used by various users in various types of servers, and then counting the usage frequency of various types of software used by various types of servers; A2. Compare the usage frequency of each type of software corresponding to each type of server with the usage frequency of the set initial test software. If the usage frequency of a certain type of software corresponding to a certain type of server is greater than or equal to the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is representative of the test. If the usage frequency of a certain type of software corresponding to a certain type of server is less than the usage frequency of the set initial test software, it means that the type of software corresponding to the type of server is not representative of the test, and the types of software corresponding to the types of servers that are representative of the test are recorded as the initial test software of each type.

3. A method for testing server hardware performance as claimed in claim 2, characterized in that: The final test software corresponding to each type of server is input into the hardware performance test platform of each type of server for performance testing. The specific test process is as follows: C1. Install each type of hardware in the target production enterprise's current cycle in each type of server into the corresponding test platform, and install the corresponding final test software of each type on the test platform of each type of server; C2. Perform a single-unit operation test on each hardware in each type of server, that is, each time the test platform runs each type of server, only one type of final test software is run on the test server, and then the single-unit performance data corresponding to the single-unit operation of each type of hardware in each type of server using the final test software is collected through the server hardware performance monitoring tool at each collection time point. The single-unit performance data includes single-unit response time, single-unit power consumption and single-unit heat dissipation efficiency; C3. After each hardware in each type of server completes the individual operation test, a combined operation test is performed on each hardware in each type of server. The combined operation test is performed according to the combined operation test of each type of final test software that appears in the final test software. That is, each time the test platform runs each type of server, only one type of final test software combination is run on the test server. Then, at each collection time point, the server hardware performance monitoring tool is used to collect the combined performance data corresponding to the combined operation of each type of hardware in each type of server using each type of final test software. The combined performance data includes combined response time, combined power consumption and combined heat dissipation efficiency.

4. A method for testing server hardware performance as claimed in claim 3, characterized in that: The analysis of the single performance evaluation values ​​of each hardware in each type of server at each collection time point in each type of final test using a single software running, the specific analysis process is as follows: The single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of final test software used in each type of server at each collection time point are obtained, and the single response time, single power consumption and single heat dissipation efficiency corresponding to the single operation of each type of final test software used in each type of server at each collection time point are input into the single performance evaluation analysis model, and the single performance evaluation value corresponding to the single operation of each type of final test software used in each type of server at each collection time point is output.

5. A method for testing server hardware performance as claimed in claim 4, characterized in that: The combined performance evaluation values ​​corresponding to the combined operation of the software combinations used in the final tests of each type of hardware in each type of server are analyzed at each collection time point. The specific analysis process is as follows: The combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the software combination running in each type of final test of each type of hardware in each type of server at each collection time point are obtained, and the combined response time, combined power consumption and combined heat dissipation efficiency corresponding to the software combination running in each type of final test of each type of hardware in each type of server at each collection time point are input into the combined performance evaluation analysis model, and the combined performance evaluation value corresponding to the software combination running in each type of final test of each type of hardware in each type of server at each collection time point is output.

6. A method for testing server hardware performance as claimed in claim 5, characterized in that: The specific analysis process of judging whether the hardware performance of each type of server is qualified is as follows: Input the single performance evaluation value corresponding to the single operation of each type of hardware in each type of server at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of final test software into the performance evaluation value evaluation model, and output the performance evaluation value results corresponding to each type of hardware in each type of server; The performance evaluation value results include values ​​of 1 and -1. When the performance evaluation value result corresponding to a certain hardware in a certain type of server is 1, it indicates that the performance of the hardware in this type of server is qualified. Conversely, when the performance evaluation value result corresponding to a certain hardware in a certain type of server is -1, it indicates that the performance of the hardware in this type of server is unqualified. If the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback will be given.

7. A method for testing server hardware performance as claimed in claim 6, characterized in that: The expression of the performance evaluation value evaluation model is: , where It represents the performance evaluation result corresponding to the r-th hardware in the q-th type server. represents the performance evaluation value corresponding to the rth hardware in the qth type server, is the set performance evaluation value threshold, q represents the number corresponding to each type of server, q is a positive integer, r represents the number corresponding to each hardware, r is a positive integer; The single performance evaluation value corresponding to the single operation of each hardware in each type of server at each collection time point and the combined performance evaluation value corresponding to the combined operation of each type of final test software are recorded as and , substitute into the calculation formula: In the above example, we can get the performance evaluation value of each hardware in each type of server. , where i represents the number corresponding to each acquisition time point, i is a positive integer, f represents the number corresponding to each type of final test using a single software run, f is a positive integer, h represents the number corresponding to each type of final test using a combination of software runs, h is a positive integer, , They are the standard single performance evaluation value corresponding to the final test of the specified hardware using the software single operation, and the standard combination performance evaluation value corresponding to the final test using the software combination operation. , They are respectively the weight factor corresponding to the single performance evaluation value of the set hardware when the software is run individually in the final test, and the weight factor corresponding to the combined performance evaluation value when the software is run in combination in the final test.

8. A method for testing server hardware performance as claimed in claim 7, characterized in that: If the performance of a certain hardware in a certain type of server is unqualified, an early warning feedback is given. The specific early warning process is as follows: When the performance evaluation results of a certain hardware in a certain type of server are monitored to be unqualified, the system will immediately activate the early warning mechanism. At this time, an early warning message will be sent to the relevant personnel responsible for server operation and maintenance. The early warning message includes the server type and hardware name.

Images