Performance evaluation method and device, equipment, storage medium and computer program product

Abstract

The invention provides a performance evaluation method and device, equipment, a storage medium and a computer program product, and is applied to a distributed block storage system. Obtaining a first data model used by the cluster and a performance value when the cluster performs load processing by using the first data model; matching the first data model with the plurality of data models to search a first static performance baseline value from the plurality of static performance baseline values when the cluster performs load processing by using the first data model; the plurality of static performance baseline values are obtained when the cluster performs load processing by using a plurality of data models under the condition that the cluster is in a no-load state; according to the performance value and the first static performance baseline value, determining a dynamic performance baseline value when the cluster performs load processing by using the first data model, so as to evaluate the performance of the distributed block storage system; the accuracy of evaluating the performance of the distributed block storage system can be improved.

Description

Technical Field

[0001] This application relates to the field of cloud storage technology, and in particular, to a performance evaluation method, device, equipment, storage medium, and computer program product. Background Art

[0002] The performance baseline of a distributed block storage system reflects the performance of the distributed block storage system under ideal conditions and is an important reference index for performance evaluation and optimization. However, the calibration of the performance baseline is a dynamic process that usually depends on manual configuration of the test environment and parameter configuration, as well as the influence of external environmental factors. The calibrated performance baseline may be inaccurate, resulting in an inability to accurately evaluate the performance of the distributed block storage system. Summary of the Invention

[0003] Embodiments of this application provide a performance evaluation method, device, equipment, storage medium, and computer program product that can dynamically calibrate the performance baseline, thereby improving the accuracy of evaluating the performance of a distributed block storage system.

[0004] To achieve the above object, the technical solution of the embodiments of this application is implemented as follows:

[0005] In a first aspect, this application proposes a performance evaluation method applied to a distributed block storage system. The method includes:

[0006] When the cluster in the distributed block storage system is under load, obtain a first data model used by the cluster and a performance value when the cluster uses the first data model for load processing;

[0007] Match the first data model with multiple data models to find a first static performance baseline value when the cluster uses the first data model for load processing from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster is under no load and the cluster uses the multiple data models for load processing;

[0008] Determine a dynamic performance baseline value when the cluster uses the first data model for load processing according to the performance value and the first static performance baseline value to evaluate the performance of the distributed block storage system.

[0009] In a second aspect, this application proposes a performance evaluation device applied to a distributed block storage system. The device includes:

[0010] An obtaining unit, configured to obtain a first data model used by the cluster and a performance value when the cluster uses the first data model for load processing when the cluster in the distributed block storage system is under load;

[0011] A search unit, configured to match the first data model with multiple data models, so as to search for a first static performance baseline value when the cluster processes the load using the first data model from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster processes the load using the multiple data models in a no-load state of the cluster.

[0012] An evaluation unit, configured to determine a dynamic performance baseline value when the cluster processes the load using the first data model according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

[0013] In a third aspect, the present application provides a performance evaluation device, including a memory, a processor, and a computer program stored on the memory and executable on the processor. When the processor executes the program, the steps of the method described in any one of the above are implemented.

[0014] In a fourth aspect, the present application provides a storage medium, on which a computer program is stored. When the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0015] In a fifth aspect, the present application provides a computer program product, including a computer program. When the computer program is executed by a processor, the steps of the method described in any one of the above are implemented.

[0016] The present application provides a performance evaluation method, device, equipment, storage medium, and computer program product, which are applied to a distributed block storage system. The method includes: when a cluster in the distributed block storage system is in a loaded state, obtaining a first data model used by the cluster and a performance value when the cluster processes the load using the first data model; matching the first data model with multiple data models, so as to search for a first static performance baseline value when the cluster processes the load using the first data model from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster processes the load using the multiple data models in a no-load state of the cluster; determining a dynamic performance baseline value when the cluster processes the load using the first data model according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system. By adopting the above implementation solution, when the distributed block storage system is in a loaded state, the first data model used by the cluster and the performance value when the cluster processes the load using the first data model are obtained in real time, and the real-time obtained performance value is combined and compared with the first static performance baseline value when the cluster processes the load using the first data model, which can truly reflect the dynamic performance baseline value when the cluster processes the load using the first data model, realize the dynamic calibration of the cluster performance baseline, and further improve the accuracy of evaluating the performance of the distributed block storage system. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 FIG. 1 is a schematic flowchart of a performance evaluation method provided by an embodiment of the present application;

[0018] Figure 2 FIG. 2 is a schematic flowchart of an exemplary static performance baseline value calibration provided by an embodiment of the present application;

[0019] Figure 3 FIG. 3 is a schematic flowchart of an exemplary dynamic performance baseline value calibration provided by an embodiment of the present application;

[0020] Figure 4 FIG. 4 is a schematic structural diagram of a performance evaluation device provided by an embodiment of the present application;

[0021] Figure 5 FIG. 5 is a schematic structural diagram of a performance evaluation device provided by an embodiment of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] In order to understand the features and technical content of the embodiments of the present application in more detail, the implementation of the embodiments of the present application will be described in detail below with reference to the accompanying drawings. The accompanying drawings are only for reference and explanation, and are not used to limit the embodiments of the present application.

[0023] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs. The terms used herein are only for the purpose of describing the embodiments of this application and are not intended to limit this application.

[0024] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict. It should also be noted that the terms "first / second" etc. involved in the embodiments of the present application are only used to distinguish similar objects and do not represent a specific order for the objects. It can be understood that "first / second" etc. can be interchanged with a specific order or sequence when allowed, so that the embodiments of the present application described here can be implemented in an order other than that illustrated or described here.

[0025] The performance baseline of a distributed block storage system refers to the lowest performance level that the system can achieve under normal working conditions. The performance baseline reflects the performance of the distributed block storage system under ideal conditions and is an important reference index for performance evaluation and optimization. Performance baseline calibration is a dynamic process. As the system evolves and upgrades, it is necessary to regularly re-evaluate and update the baseline standard to ensure that the performance of the system is always in the best state.

[0026] The performance baseline of a distributed block storage system usually includes throughput, response time, concurrent performance, Input / Output Operations Per Second (IOPS), latency, bandwidth, reliability, availability, scalability, etc. Usually, the determination of the performance baseline of a distributed block storage system needs to be obtained through actual testing and evaluation. This involves conducting load testing, stress testing, and performance testing on the system, etc., to simulate the usage in a real scenario, and determining the performance baseline of the system based on the test results. According to the performance baseline, performance optimization and adjustment can be further carried out to improve the performance and reliability of the system.

[0027] Common performance baseline calibration methods for existing distributed block storage systems include stress testing, throughput testing, response time testing, scalability testing, fault recovery testing, etc. However, there are the following disadvantages:

[0028] (1) Lack of test automation ability. Currently, the performance calibration of distributed block storage systems usually relies on manual configuration of the test environment and parameter settings. This method is easily affected by subjective factors, and the test results may not be accurate and repeatable. At the same time, a distributed block storage system usually consists of multiple nodes, involving complex mechanisms and algorithms such as data distribution, data consistency, and load balancing. This makes the test process more complex and requires considering the coordination and interaction between each node.

[0029] (2) Lack of consistency and repeatability. The performance of a distributed block storage system is often affected by various factors, including network latency, node load, data scale, etc. These factors may change at different times and in different environments, resulting in unstable calibration results of the performance baseline. Due to the inconsistency of the test environment and parameter configuration, the results of the same performance test may vary at different times and in different environments. The lack of consistency and repeatability affects the credibility of the performance calibration results to a certain extent.

[0030] (3) Difficulty in covering diverse usage scenarios. Due to the complexity and variability of distributed block storage systems, distributed block storage systems usually face diverse usage scenarios and load patterns, such as read-write ratio, data distribution, access mode, Input / Output (IO) model, etc. The cluster performance is a dynamic value, and it is difficult to find a test scenario that fully represents all usage cases and load patterns of the system. Existing performance calibration methods usually target a fixed range of test models in a laboratory environment to calibrate clear performance baseline values, resulting in the evaluation results may differ from the actual usage.

[0031] (4) Difficulties in the selection and management of test data. Test data is required to cover various data types, access patterns, and load conditions to accurately evaluate the performance of the system. Performance calibration needs to use appropriate test data for testing, but selecting appropriate test data and managing its scale and distribution is a challenge. Existing methods often struggle to find test data that fully represents the characteristics of actual data.

[0032] (5) Difficulties in evaluating the scalability and elasticity of the system. Performance calibration of a distributed block storage system usually needs to consider the scalability and elasticity of the system. However, existing performance calibration methods often struggle to accurately evaluate the scalability and elasticity performance of the system under different scales and loads.

[0033] (6) System performance baseline.

[0034] Based on this, an embodiment of the present application provides a performance evaluation method. Figure 1 The following is a schematic flowchart of a performance evaluation method provided by an embodiment of the present application; as Figure 1 shown, applied to a distributed block storage system, the method includes:

[0035] S101. When the cluster in the distributed block storage system is under load, obtain the first data model used by the cluster and the performance value when the cluster uses the first data model for load processing.

[0036] It should be noted that the first data model can be understood as the business data model used by the cluster, and this business data model can be any data model, which can be specifically determined according to the actual situation and is not limited here. The performance value can be understood as the real-time performance value when the cluster uses the first data model for load processing; that is, the process of obtaining the first data model and the performance value is obtained in real time. Obtaining the first data model used by the cluster and the performance value when the cluster uses the first data model for load processing can be understood as that the distributed block storage system enters the cluster data model acquisition module, collects the cluster business data model in real time according to this cluster data module acquisition model, and obtains the cluster real-time performance value.

[0037] S102. Match the first data model with multiple data models to find the first static performance baseline value when the cluster uses the first data model for load processing from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster is under no load and the cluster uses multiple data models for load processing.

[0038] It should be noted that the multiple data models can be understood as multiple preset data models. The types and quantities of the multiple data models can be determined according to the actual situation and are not limited herein. As an example, the multiple data models can be exemplified as common data models of distributed block storage with the default configuration of the cluster calibration system, such as data models of 4 Kilobytes (KB), 32 KB, 64 KB, 128 KB, 256 KB, 512 KB, 1 Mbyte (MB), and 4 MB. The multiple static performance baseline values can be understood as the maximum performance baseline values when the cluster processes the load using the multiple data models under no-load conditions.

[0039] In the embodiment of the present application, the process of obtaining the multiple static performance baseline values specifically includes: when the cluster is under no-load, using the multiple data models to apply load pressure to the cluster respectively; during the process of applying load pressure to the cluster, performing dial test processing on the cluster to obtain multiple dial test results when the cluster processes the load using the multiple data models; and determining the multiple static performance baseline values according to the multiple dial test results and the preset performance baseline values.

[0040] It should be noted that the static performance baseline values are determined within the cluster performance baseline calibration system. Using the multiple data models to apply load pressure to the cluster respectively can be understood as applying load pressure to the cluster with each of the multiple data models respectively, and the process of applying load pressure to the cluster with each data model is continuous.

[0041] It should be noted that using the multiple data models to apply load pressure to the cluster respectively; during the process of applying load pressure to the cluster, performing dial test processing on the cluster to obtain multiple dial test results when the cluster processes the load using the multiple data models can be understood as, after the cluster starts the dial test processing, using the multiple data models to apply load pressure to the cluster respectively and simultaneously observing the multiple dial test results output by the cluster during the process of applying load pressure to the cluster, and each dial test result corresponds to each data model one by one.

[0042] In the embodiment of the present application, the preset performance baseline value can be understood as the single-instance performance baseline value promised by the distributed block storage system product to the customer. This single-instance performance baseline value promised to the customer can be used as a key parameter to measure the performance water level of the cluster. The single-instance performance baseline value promised to the customer includes two parts. One part is the preset latency information, and the other part is the number of input / output operations per second (Input / Output Operations Per Second, IOPS). The process of determining multiple static performance baseline values based on multiple dial test results and the preset performance baseline value specifically includes: determining the first latency information corresponding to the first dial test result among the multiple dial test results, and obtaining the second latency information in the preset performance baseline value; when the difference between the first latency information and the second latency information is less than the first threshold, searching for the second data model corresponding to the first dial test result from multiple data models, and using the pressure load value under the second data model as the static performance baseline value corresponding to the cluster when using the second data model for load processing.

[0043] It should be noted that the first dial test result can be any one of the multiple dial test results. The first latency information corresponding to the first dial test result can be understood as the real-time latency information output during the process of applying pressure load to the second data model after the cluster starts the dial test. The second latency information can be understood as the preset latency information in the single-instance baseline value promised to the customer. The difference between the first latency information and the second latency information can be understood as: when the first latency information is greater than the second latency information, the difference obtained by subtracting the second latency information from the first latency information; when the first latency information is less than or equal to the second latency information, the difference obtained by subtracting the first latency information from the second latency information. The value of the first threshold can be determined according to the actual situation and is not limited here. As an example, the first threshold can be 0.5%. It should be noted that searching for the second data model corresponding to the first dial test result from multiple data models and using the pressure load value under the second data model as the static performance baseline value corresponding to the cluster when using the second data model for load processing can be understood as searching for the second data model corresponding to the first dial test result from multiple data models and using the current load pressure value of the second data model as the static performance baseline value of the cluster when using the second data model for load processing. Among them, the load pressure value at this time can be understood as the current IOPS of the data model.

[0044] For the convenience of understanding, in practical applications, it can be exemplified that when the first delay information is greater than the second delay information in the performance baseline value of the single instance promised to the customer, and the difference between the first delay information and the second delay information in the performance baseline value of the single instance promised to the customer is less than or equal to 0.5%, the pressure load value of the second data model corresponding to the first test result at this time is saved, and this pressure load value is used as the static performance baseline value when the cluster uses the second data model for load processing; or, when the first delay information is less than or equal to the second delay information in the performance baseline value of the single instance promised to the customer, and the difference between the first delay information and the second delay information in the performance baseline value of the single instance promised to the customer is less than or equal to 0.5%, the pressure load value of the second data model corresponding to the first test result at this time is saved, and this pressure load value is used as the static performance baseline value when the cluster uses the second data model for load processing.

[0045] In the embodiment of the present application, the method further includes: determining the first delay information corresponding to the first test result among multiple test results, and obtaining the second delay information in the preset performance baseline value; when the first delay information is greater than the second delay information and the difference between the first delay information and the second delay information is greater than the first threshold, performing load decompression on the second data model corresponding to the first test result; when the first delay information is less than or equal to the second delay information and the difference between the first delay information and the second delay information is greater than the first threshold, performing load pressurization on the second data model.

[0046] It should be noted that the meanings of the first test result, the first delay information, the preset performance baseline value, the second delay information, and the first threshold here are the same as those above, and will not be elaborated here.

[0047] It should be noted that during the process of performing load decompression on the second data model corresponding to the first test result, it is necessary to collect the test results of the cluster in real time until the difference between the delay information corresponding to the real-time test result of the second data model and the second delay information in the preset performance baseline value is less than or equal to the first threshold, then stop performing load decompression on the second data model, obtain the current pressure load value of the second data model, and use this pressure load value as the static performance baseline value when the cluster uses the second data model for load processing. Among them, in practical applications, performing load decompression on the second data model can be understood as reducing the IOPS of the second data model.

[0048] It should be noted that during the process of applying load pressure to the second data model corresponding to the first batch test result, it is necessary to collect the batch test results of the cluster in real time until the difference between the latency information corresponding to the real-time batch test result of the second data model and the second latency information in the preset performance baseline value is less than or equal to the first threshold. Then, stop applying load pressure to the second data model, obtain the current pressure load value of the second data model, and use this pressure load value as the static performance baseline value when the cluster uses the second data model for load processing. Among them, in practical applications, applying load pressure to the second data model can be understood as increasing the IOPS of the second data model.

[0049] For ease of understanding, an example is given here. When the first latency information is greater than the second latency information in the performance baseline value of a single instance promised to customers and the difference between the first latency information and the second latency information in the performance baseline value of a single instance promised to customers is greater than 0.5%, perform load decompression on the second data model corresponding to the first batch test result; when the first latency information is less than or equal to the second latency information in the performance baseline value of a single instance promised to customers and the difference between the first latency information and the second latency information in the performance baseline value of a single instance promised to customers is greater than 0.5%, perform load pressure on the second data model corresponding to the first batch test result.

[0050] It should be noted that each data model in multiple data models needs to obtain the corresponding static performance baseline value according to the batch test result and the performance baseline value promised to customers.

[0051] The solution of the embodiment of the present application, by introducing the single instance batch test function as a load pressure measurement tool, shields the differences in the cluster environment, architecture, hardware, etc. Without manual intervention, it can automatically complete the calibration of the static performance baseline value of the distributed block storage system in a multi-model scenario.

[0052] In the embodiment of the present application, matching the first data model with multiple data models to find the first static performance baseline value when the cluster uses the first data model for load processing can be understood as matching the first data model collected in real time with multiple data models to obtain the corresponding first static performance baseline value when the cluster uses the first data model for load processing.

[0053] S103. Determine the dynamic performance baseline value when the cluster uses the first data model for load processing according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

[0054] In the embodiment of the present application, the dynamic performance baseline value can be understood as the cluster performance water level value. The process of determining the dynamic performance baseline value when the cluster uses the first data model for load processing according to the performance value and the first static performance baseline value specifically includes: performing a dial test on the cluster to obtain a second dial test result when the cluster uses the first data model for load processing; updating the first static performance baseline value according to the second dial test result and the performance value to obtain a second static performance baseline value; and determining the dynamic performance baseline value according to the performance value and the second static performance baseline value.

[0055] It should be noted that the second dial test result can be understood as the dial test result output when the cluster uses the first data model for load processing under the condition of being loaded.

[0056] In the embodiment of the present application, the second static performance baseline value can be understood as the performance baseline value obtained after updating the first static performance baseline value. The process of updating the first static performance baseline value according to the second dial test result and the performance value to obtain a second static performance baseline value specifically includes: when the performance value is greater than the first static performance baseline value, determining the third delay information corresponding to the second dial test result and obtaining the second delay information in the preset performance baseline value; when the ratio between the third delay information and the second delay information is greater than the second threshold, taking the first static performance baseline value as the second static performance baseline value; and when the ratio between the third delay information and the second delay information is less than or equal to the second threshold, updating the first static performance baseline value to the second static performance baseline value according to the performance value.

[0057] It should be noted that the preset performance baseline value here can also be understood as the performance baseline value of a single instance committed to the customer. As a key parameter for measuring the cluster performance water level, in practical applications, the performance baseline value of a single instance committed to the customer includes two parts, one part is the preset delay information, and the other part is the IOPS. The third delay information can be understood as the real-time delay information during the actual operation of the first data model after the cluster starts the dial test. The meaning of the second delay information has been described above and will not be repeated here. The value of the second threshold can be determined according to the actual situation and is not limited here. As an example, the second threshold can be set to 95%. The ratio between the third delay information and the second delay information being greater than the second threshold can be exemplified as the ratio between the third delay information and the second delay information being greater than 95%, or it can also be exemplified as the third delay information being greater than 95% of the second delay information.

[0058] It should be noted that taking the first static performance baseline value as the second static performance baseline value can be understood as that the second static performance baseline value is the same as the first static performance baseline value. In the case where the performance value is greater than the first static performance baseline value and the ratio between the third delay information and the second delay information in the preset performance baseline value is greater than the second threshold, taking the first static performance baseline value as the second static performance baseline value can be exemplified as follows: in the case where the real-time performance value is greater than the first static performance baseline value and the third delay information is greater than 95% of the second delay information in the performance baseline value for a single instance promised to the customer, there is no need to update the static performance baseline value, that is, taking the first static performance baseline value as the second static performance baseline value.

[0059] It should be noted that updating the first static performance baseline value to the second static performance baseline value according to the performance value, and updating the first static performance baseline value to the second static performance baseline value according to the real-time performance value means that at this time the second static performance baseline value is the same as the real-time performance value. In the case where the performance value is greater than the first static performance baseline value and the ratio between the third delay information and the second delay information in the preset performance baseline value is less than or equal to the second threshold, updating the first static performance baseline value to the second static performance baseline value according to the performance value can be exemplified as follows: in the case where the real-time performance value is greater than the first static performance baseline value and the third delay information is less than or equal to 95% of the second delay information in the performance baseline value for a single instance promised to the customer, then update the first static performance baseline value to the second static performance baseline value with the real-time performance value.

[0060] In the embodiments of the present application, the process of updating the first static performance baseline value to obtain the second static performance baseline value according to the second dial test result and the performance value specifically includes: in the case where the performance value is less than or equal to the first static performance baseline value, determining the third delay information corresponding to the second dial test result, and obtaining the second delay information in the preset performance baseline value; in the case where the ratio between the third delay information and the second delay information is greater than the third threshold, outputting an alarm signal to check whether the cluster is abnormal; in the case where the cluster is not abnormal, updating the first static performance baseline value to the second static performance baseline value according to the performance value. It should be noted that the meanings of the third delay information and the second delay information have been described above and will not be elaborated here. The value of the third threshold can be determined according to the actual situation and is not limited here. As an example, the third threshold can be 105%. The ratio between the third delay information and the second delay information being greater than the third threshold can be exemplified as follows: the ratio between the third delay information and the second delay information is greater than 105%, which can also be understood as the third delay information being greater than 105% of the second delay information. The alarm signal can be understood as a cluster alarm signal.

[0061] For the convenience of understanding, an example is given here. When the real-time performance value is less than or equal to the first static performance baseline value, and the ratio between the third latency information and the second latency information in the performance baseline value for a single instance of the commitment to customers is greater than 105%, a cluster alarm signal is output to check whether the cluster is abnormal. If the cluster is not abnormal, the first static performance baseline value is updated to the second static performance baseline value with the real-time performance value.

[0062] In the embodiment of the present application, after checking whether the cluster is abnormal, the method further includes: when the cluster is abnormal, performing fault handling on the cluster.

[0063] It should be noted that when the cluster is abnormal, corresponding fault handling is performed on the cluster.

[0064] In the embodiment of the present application, the process of updating the first static performance baseline value to obtain the second static performance baseline value according to the second dial test result and the performance value further includes: when the performance value is less than or equal to the first static performance baseline value, determining the third latency information corresponding to the second dial test result, and obtaining the second latency information in the preset performance baseline value; when the ratio between the third latency information and the second latency information is less than or equal to the third threshold and greater than the second threshold, updating the first static performance baseline value to the second static performance baseline value according to the performance value; when the ratio between the third latency information and the second latency information is less than or equal to the second threshold, using the first static performance baseline value as the second static performance baseline value.

[0065] It should be noted that the meanings of the third latency information and the second latency information have been described above and will not be repeated here. The values of the second threshold and the third threshold are the same as those above. The second threshold can be exemplified as 95%, and the third threshold can be exemplified as 105%. The ratio between the third latency information and the second latency information being less than or equal to the third threshold and greater than the second threshold can be understood as the ratio between the third latency information and the second latency information being less than or equal to 105% and greater than 95%, or it can also be understood as the third latency information being less than or equal to 105% of the second latency information but greater than 95% of the second latency information.

[0066] It should be noted that in the case where the performance value is less than or equal to the first static performance baseline value, the ratio between the third delay information and the second delay information is less than or equal to the third threshold and greater than the second threshold, updating the first static performance baseline value to the second static performance baseline value according to the performance value can be exemplified as follows: when the real-time performance value is less than or equal to the first static performance baseline value and the third delay information is less than or equal to 105% of the second delay information in the performance baseline value for a single instance promised to the customer, but greater than 95% of the second delay information in the performance baseline value for a single instance promised to the customer, updating the first static performance baseline value to the second static performance baseline value according to the real-time performance value.

[0067] It should be noted that the ratio between the third delay information and the second delay information being less than or equal to the second threshold can be understood as the ratio between the third delay information and the second delay information in the performance baseline value for a single instance promised to the customer being less than or equal to 95%, or it can also be understood as the third delay information being less than or equal to 95% of the second delay information in the performance baseline value for a single instance promised to the customer. In the case where the performance value is less than or equal to the first static performance baseline value and the ratio between the third delay information and the second delay information in the preset performance baseline value is less than or equal to the second threshold, taking the first static performance baseline value as the second static performance baseline value can be exemplified as follows: when the real-time performance value is less than or equal to the first static performance baseline value and the third delay information is less than or equal to 95% of the second delay information in the performance baseline value for a single instance promised to the customer, there is no need to update the first static performance baseline value, and directly taking the first static performance baseline value as the second static performance baseline value.

[0068] In the embodiment of the present application, determining the dynamic performance baseline value according to the performance value and the second static performance baseline value can be understood as taking the ratio of the performance value and the second static performance baseline value as the dynamic performance baseline value.

[0069] It should be noted that the determination of the above static performance baseline value and dynamic performance baseline value can also be understood as calibrating the static performance baseline value and dynamic performance baseline value, and its calibration process can also be understood as being carried out under a calibration system. After determining the dynamic performance baseline value, judge whether the system ends according to the system stop condition set by the calibration system. The stop condition is set as the running duration of the calibration system, the cluster performance water level value, etc.

[0070] The solution of the embodiment of the present application, by introducing the real-time acquisition function of the cluster data model and combining with the cluster static performance baseline value, shields the test model design problems brought by customer diversity, service diversity, etc. in the public cloud scenario, and realizes the dynamic calibration of the cluster performance baseline by introducing the performance baseline dynamic compensation value system.

[0071] The solution of the embodiment of the present application pairs the obtained real business data model with the performance baseline of the corresponding data model of the cluster performance baseline, which can truly reflect the performance level of the cluster at this time point and under this business model, and realizes the dynamic calculation and display of the cluster performance level under different data models.

[0072] For ease of understanding, the determination processes of the above-mentioned static performance baseline value and dynamic performance baseline value are illustrated by examples. Specifically, the static performance baseline value is determined within the cluster static performance baseline calibration system module, and the cluster dynamic performance baseline value is determined within the cluster dynamic performance baseline calibration system module.

[0073] 1. Cluster static performance baseline calibration system module. It is mainly used for calibrating the cluster performance baseline under the condition of the cluster being idle. The calibration result output by the static performance baseline calibration system module serves as the base value for the dynamic performance baseline calibration system module to adjust the performance baseline under the condition of the cluster carrying business load, avoiding the dynamic performance baseline calibration system module blindly applying pressure under the condition of the cluster carrying load and thus affecting the cluster business. During the cluster performance baseline calibration process, the system continuously applies load pressure to the cluster with a fixed data model sample. By introducing the single-instance probing function, within the range of the promised performance to customers, the maximum performance baseline value of the cluster under a specific data model sample is calculated, and thus the calibration of the static performance baseline value of the cluster is completed. Figure 2 It is a schematic flow diagram of an exemplary static performance baseline value calibration provided by the embodiment of the present application; as Figure 2 shown, the specific steps are as follows:

[0074] 1. Start.

[0075] 2. Input the promised single-instance performance baseline to customers.

[0076] 3. Start probing.

[0077] 4. Whether to enter the cluster static performance baseline calibration.

[0078] It should be noted that in the case of entering the cluster static performance baseline calibration, step 5 is executed.

[0079] 5. Apply pressure load with multiple data models.

[0080] It should be noted that the cluster is continuously subjected to pressure load using multiple data models.

[0081] 6. Determine whether the probing result exceeds the promised single-instance performance baseline to customers.

[0082] It should be noted that when the dial test result exceeds the performance baseline of a single instance in the commitment form to the customer, step 7 is executed; when the dial test result does not exceed the performance baseline of a single instance in the commitment form to the customer, step 8 is executed.

[0083] 7. Determine whether the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer exceeds 0.5%.

[0084] It should be noted that when the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer exceeds 0.5%, step 9 is executed; when the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer does not exceed 0.5%, step 10 is executed.

[0085] 8. Determine whether the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer exceeds 0.5%.

[0086] It should be noted that when the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer exceeds 0.5%, step 11 is executed; when the deviation between the dial test result and the performance baseline of a single instance in the commitment form to the customer does not exceed 0.5%, step 10 is executed.

[0087] 9. Decompress with the same data model.

[0088] 10. Save the performance baseline.

[0089] It should be noted that the load pressure under the current data model is saved as the static performance baseline value under this data model. After the performance baseline is saved, step 12 is executed.

[0090] 11. Continuously pressurize with the same data model.

[0091] 12. End.

[0092] For the convenience of understanding, the above steps are described in detail here:

[0093] (1) Input the performance baseline of a single instance in the commitment form to the customer of the distributed block storage system product as the key parameter to measure the performance water level of the cluster;

[0094] (2) Start the cluster dial test instance, and the output dial test result is used as the key indicator to measure the availability of the cluster.

[0095] (3) Determine whether to enter the cluster static performance baseline calibration system module according to the system parameters.

[0096] (4) In step (3), if entering the cluster performance baseline calibration system, static performance baseline calibration is performed serially according to the common data models of distributed block storage configured by default in the system, such as 4KB, 32KB, 64KB, 128KB, 256KB, 512KB, 1MB, 4MB block size sub-model scenarios.

[0097] (5) During the process of loading the load, synchronously observe the measurement results of the measurement system. If the measurement result exceeds the performance baseline value promised to the customer: If the deviation from the performance baseline value promised to the customer does not exceed 0.5%, then save the load pressure under the current data model as the static performance baseline under this data model; If the deviation from the performance baseline value promised to the customer exceeds 0.5%, then reduce the load on the current data model. If the measurement result does not exceed the performance baseline value promised to the customer: If the deviation from the performance baseline value promised to the customer does not exceed 0.5%, then save the load pressure under the current data model as the static performance baseline under this data model; If the deviation from the performance baseline value promised to the customer exceeds 0.5%, then increase the load on the current data model.

[0098] (6) When the static performance baseline calibration results of all data models are output, the static performance calibration system stops.

[0099] 2. Cluster dynamic performance baseline calibration system module. It is mainly used for calibrating the cluster performance baseline under the condition of the cluster being loaded. The cluster dynamic performance baseline calibration module is based on the real-time service model, refers to the cluster static performance baseline value, and through introducing single-instance performance measurement, within the performance range promised to the customer, completes the dynamic adjustment of the cluster performance baseline, and finally outputs the cluster performance baseline value of the cluster under the real business scenario, service model, and data I / O model. Figure 3 A schematic flow chart of an exemplary dynamic performance baseline value calibration provided by an embodiment of this application; as Figure 3 shown, the specific steps are as follows:

[0100] 1. Start.

[0101] 2. Input the single-instance performance baseline promised to the customer.

[0102] 3. Start the measurement.

[0103] 4. Whether to enter the cluster static performance baseline calibration.

[0104] It should be noted that in the case of entering the cluster static performance baseline calibration, execute step 5.

[0105] 5. Real-time collect the cluster data model.

[0106] 6. Obtain the performance value of the data model corresponding to the cluster static performance baseline.

[0107] It should be noted that match the cluster data model with the data model in the cluster static performance baseline to obtain the cluster static performance baseline calibration value corresponding to the current cluster service data model.

[0108] 7. Whether the real-time cluster performance value exceeds the static performance baseline value.

[0109] It should be noted that when the real-time performance value of the cluster exceeds the static performance baseline value, step 8 is executed; when the real-time performance value of the cluster does not exceed the static performance baseline value, step 11 is executed.

[0110] 8. Whether the test result exceeds 95% of the performance baseline of a single instance promised to the customer.

[0111] It should be noted that when the test result does not exceed 95% of the performance baseline of a single instance promised to the customer, step 9 is executed; when the test result exceeds 95% of the performance baseline of a single instance promised to the customer, step 10 is executed.

[0112] 9. Update the performance value of the data model corresponding to the static performance value of the cluster with the real-time performance value.

[0113] It should be noted that step 10 is executed after step 9.

[0114] 10. Output the cluster performance water level value.

[0115] It should be noted that the ratio between the real-time performance value and the current static performance baseline value is used as the cluster performance water level value; after step 10, step 16 is executed.

[0116] 11. Whether the test result exceeds 105% of the performance baseline of a single instance promised to the customer.

[0117] It should be noted that when the test result exceeds 105% of the performance baseline of a single instance promised to the customer, step 12 is executed; when the test result does not exceed 105% of the performance baseline of a single instance promised to the customer, step 15 is executed.

[0118] 12. Trigger an alarm and check whether the cluster is abnormal.

[0119] 13. Whether the cluster is abnormal.

[0120] When the cluster is abnormal, step 14 is executed; when the cluster is not abnormal, step 9 is executed.

[0121] 14. Complete the fault handling.

[0122] 15. Whether the test result exceeds 95% of the performance baseline of a single instance promised to the customer.

[0123] It should be noted that when the test result exceeds 95% of the performance baseline of a single instance promised to the customer, step 9 is executed; when the test result does not exceed 95% of the performance baseline of a single instance promised to the customer, step 10 is executed.

[0124] 16. Whether the dynamic performance measurement stop condition is reached.

[0125] It should be noted that step 5 is executed when the dynamic performance measurement stop condition is not reached; step 17 is executed when the dynamic performance measurement stop condition is reached.

[0126] 17. End.

[0127] For the convenience of understanding, the above steps are described in detail as follows:

[0128] (1) Input the instance performance baseline of the distributed block storage system product's commitment to customers as a key parameter for measuring the cluster performance water level.

[0129] (2) Start the cluster dial test instance, and the output dial test result is used as a key indicator for measuring the cluster availability.

[0130] (3) Determine whether to enter the cluster dynamic performance baseline calibration system according to the system parameters.

[0131] (4) Enter the cluster data model acquisition module to collect the cluster business data model in real time. The collected data model is matched with the data model in the cluster static performance baseline to obtain the cluster static performance baseline calibration value corresponding to the current cluster business data model.

[0132] (5) Obtain the cluster real-time performance value and compare it with the cluster static performance baseline calibration value (hereinafter referred to as the static performance baseline value) corresponding to the current business data model.

[0133] (6) If the real-time performance value exceeds the static performance baseline value:

[0134] A. If the dial test result exceeds 95% of the commitment baseline to customers, there is no need to update the static performance baseline.

[0135] B. If the dial test result does not exceed 95% of the commitment baseline to customers, then update the calibration value of the cluster static performance baseline corresponding to the current business data model with the real-time performance value under the current business data model.

[0136] (7) If the real-time performance value does not exceed the static performance baseline value:

[0137] A. If the dial test result exceeds 105% of the commitment baseline to customers, output a cluster alarm and check whether the cluster is abnormal. If the cluster is abnormal, perform corresponding fault handling; if the cluster is not abnormal, update the calibration value of the cluster static performance baseline corresponding to the current business data model with the real-time performance value under the current business data model.

[0138] B. If the dial test result does not exceed 105% of the commitment baseline to customers but exceeds 95% of the commitment baseline to customers, then update the calibration value of the cluster static performance baseline corresponding to the current business data model with the real-time performance value under the current business data model.

[0139] C. If the test result does not exceed 95% of the baseline committed to customers, there is no need to update the static performance baseline.

[0140] (8) Use the ratio of the current real-time performance value to the real-time performance value under the current service data model as the cluster performance water level value for operation and maintenance reference.

[0141] (9) Determine whether the system ends according to the system stop conditions set by the dynamic performance baseline calibration system. The stop conditions are set as the operation duration of the calibration system, the cluster performance water level value, etc.

[0142] The performance baseline output by the dynamic performance baseline calibration system module is the cluster performance baseline value under the current real business scenario, business model, and data I / O model.

[0143] The embodiment of the present application provides a performance evaluation device. Figure 4 It is a schematic structural diagram of a performance evaluation device provided by the embodiment of the present application; as Figure 4 shown, it is applied to a distributed block storage system. The performance evaluation device 400 includes:

[0144] An acquisition unit 401, configured to acquire a first data model used by the cluster and a performance value when the cluster processes loads using the first data model when the cluster in the distributed block storage system is under load.

[0145] A search unit 402, configured to match the first data model with multiple data models to search for a first static performance baseline value when the cluster processes loads using the first data model from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster processes loads using the multiple data models when the cluster is idle.

[0146] An evaluation unit 403, configured to determine a dynamic performance baseline value when the cluster processes loads using the first data model according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

[0147] Optionally, the performance evaluation device 400 further includes a determination unit, configured to, when the cluster is idle, apply load pressure to the cluster respectively using the multiple data models; when the cluster is under load pressure, perform a test on the cluster to obtain multiple test results when the cluster processes loads using the multiple data models; and determine the multiple static performance baseline values according to the multiple test results and a preset performance baseline value.

[0148] Optionally, the determining unit is further configured to determine first delay information corresponding to a first test result among the multiple test results, and obtain second delay information in the preset performance baseline value; when the difference between the first delay information and the second delay information is less than a first threshold, search for a second data model corresponding to the first test result from the multiple data models, and use the pressure load value under the second data model as the static performance baseline value corresponding to the cluster when using the second data model for load processing.

[0149] Optionally, the performance evaluation device 400 further includes a load processing unit, configured to determine first delay information corresponding to a first test result among the multiple test results, and obtain second delay information in the preset performance baseline value; when the first delay information is greater than the second delay information and the difference between the first delay information and the second delay information is greater than a first threshold, perform load decompression on the second data model corresponding to the first test result; when the first delay information is less than or equal to the second delay information and the difference between the first delay information and the second delay information is greater than a first threshold, perform load pressurization on the second data model.

[0150] Optionally, the evaluation unit 403 is further configured to perform a test on the cluster to obtain a second test result when the cluster uses the first data model for load processing; update the first static performance baseline value according to the second test result and the performance value to obtain a second static performance baseline value; determine the dynamic performance baseline value according to the performance value and the second static performance baseline value.

[0151] Optionally, the evaluation unit 403 is further configured to, when the performance value is greater than the first static performance baseline value, determine third delay information corresponding to the second test result, and obtain second delay information in the preset performance baseline value; when the ratio between the third delay information and the second delay information is greater than a second threshold, use the first static performance baseline value as the second static performance baseline value; when the ratio between the third delay information and the second delay information is less than or equal to the second threshold, update the first static performance baseline value to the second static performance baseline value according to the performance value.

[0152] Optionally, the evaluation unit 403 is further configured to determine third delay information corresponding to the second dial test result and obtain second delay information in a preset performance baseline value when the performance value is less than or equal to the first static performance baseline value; output an alarm signal to check whether the cluster is abnormal when a ratio between the third delay information and the second delay information is greater than a third threshold; and update the first static performance baseline value to the second static performance baseline value according to the performance value when the cluster is normal.

[0153] Optionally, after checking whether the cluster is abnormal, the performance evaluation device 400 further includes a fault handling unit configured to handle a fault of the cluster when the cluster is abnormal.

[0154] Optionally, the evaluation unit 403 is further configured to determine third delay information corresponding to the second dial test result and obtain second delay information in a preset performance baseline value when the performance value is less than or equal to the first static performance baseline value; update the first static performance baseline value to the second static performance baseline value according to the performance value when a ratio between the third delay information and the second delay information is less than or equal to the third threshold and greater than a second threshold; and use the first static performance baseline value as the second static performance baseline value when the ratio between the third delay information and the second delay information is less than or equal to the second threshold. An embodiment of the present application further provides a performance evaluation device. Figure 5 is a schematic structural diagram of a performance evaluation device provided by an embodiment of the present application; as Figure 5 shown, the performance evaluation device 500 includes: a processor 501 and a memory 503. Optionally, the performance evaluation device may further include a communication bus 502.

[0155] During the process of a specific embodiment, the above-mentioned processor 501 may be at least one of an application specific integrated circuit (ASIC), a digital signal processor (DSP), a digital signal processing image processing device (DSPD), a programmable logic image processing device (PLD), a field programmable gate array (FPGA), a CPU, a controller, a microcontroller, and a microprocessor. It can be understood that for different devices, the electronic devices used to implement the functions of the above-mentioned processor may also be other, and this embodiment does not make specific limitations.

[0156] In the embodiment of the present application, the above-mentioned communication bus 502 is used to realize the connection and communication between the processor 501 and the memory 503; when the above-mentioned processor 501 executes the running program stored in the memory 503, the following performance evaluation method is realized:

[0157] When the cluster in the distributed block storage system is under load, obtain the first data model used by the cluster and the performance value when the cluster uses the first data model for load processing; match the first data model with multiple data models to find the first static performance baseline value when the cluster uses the first data model for load processing from multiple static performance baseline values; the multiple static performance baseline values are obtained when the cluster uses the multiple data models for load processing when the cluster is idle; determine the dynamic performance baseline value when the cluster uses the first data model for load processing according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

[0158] Furthermore, the above-mentioned processor 501 is further used to, when the cluster is idle, use the multiple data models to respectively apply load pressure to the cluster; when the cluster is under load pressure, perform dial test processing on the cluster to obtain multiple dial test results when the cluster uses the multiple data models for load processing; determine the multiple static performance baseline values according to the multiple dial test results and the preset performance baseline value.

[0159] Further, the above-mentioned processor 501 is further configured to determine first delay information corresponding to a first test result among the multiple test results, and obtain second delay information in the preset performance baseline value; when the difference between the first delay information and the second delay information is less than a first threshold, search for a second data model corresponding to the first test result from the multiple data models, and use the pressure load value under the second data model as the static performance baseline value corresponding to the cluster when using the second data model for load processing.

[0160] Further, the above-mentioned processor 501 is further configured to determine first delay information corresponding to a first test result among the multiple test results, and obtain second delay information in the preset performance baseline value; when the first delay information is greater than the second delay information and the difference between the first delay information and the second delay information is greater than a first threshold, perform load decompression on the second data model corresponding to the first test result; when the first delay information is less than or equal to the second delay information and the difference between the first delay information and the second delay information is greater than a first threshold, perform load pressurization on the second data model.

[0161] Further, the above-mentioned processor 501 is further configured to perform a test on the cluster to obtain a second test result when the cluster uses the first data model for load processing; update the first static performance baseline value according to the second test result and the performance value to obtain a second static performance baseline value; determine the dynamic performance baseline value according to the performance value and the second static performance baseline value.

[0162] Further, the above-mentioned processor 501 is further configured to, when the performance value is greater than the first static performance baseline value, determine third delay information corresponding to the second test result, and obtain second delay information in the preset performance baseline value; when the ratio between the third delay information and the second delay information is greater than a second threshold, use the first static performance baseline value as the second static performance baseline value; when the ratio between the third delay information and the second delay information is less than or equal to the second threshold, update the first static performance baseline value to the second static performance baseline value according to the performance value.

[0163] Further, the above-mentioned processor 501 is further configured to, when the performance value is less than or equal to the first static performance baseline value, determine third delay information corresponding to the second dial test result, and obtain second delay information in the preset performance baseline value; when the ratio between the third delay information and the second delay information is greater than a third threshold, output an alarm signal to check whether the cluster has an abnormality; when the cluster has no abnormality, update the first static performance baseline value to the second static performance baseline value according to the performance value.

[0164] Further, after checking whether the cluster has an abnormality, the above-mentioned processor 501 is further configured to, when the cluster has an abnormality, perform fault handling on the cluster.

[0165] Further, the above-mentioned processor 501 is further configured to, when the performance value is less than or equal to the first static performance baseline value, determine third delay information corresponding to the second dial test result, and obtain second delay information in the preset performance baseline value; when the ratio between the third delay information and the second delay information is less than or equal to the third threshold and greater than a second threshold, update the first static performance baseline value to the second static performance baseline value according to the performance value; when the ratio between the third delay information and the second delay information is less than or equal to the second threshold, use the first static performance baseline value as the second static performance baseline value.

[0166] An embodiment of the present application provides a storage medium, on which a computer program is stored. The above-mentioned computer-readable storage medium stores one or more programs, and the one or more programs can be executed by one or more processors, and the computer program implements the performance evaluation method as described above.

[0167] Based on the above embodiments, an embodiment of the present application provides a computer program product, including a computer program, where the computer program can be executed by one or more processors, and the computer program implements the performance evaluation method as described above.

[0168] It should be noted that in this article, the terms "include", "comprise" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "including a..." does not exclude the existence of another identical element in the process, method, article or device including the element.

[0169] Through the description of the above embodiments, those skilled in the art can clearly understand that the above-described example methods can be implemented by means of software plus a necessary general hardware platform. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation. Based on such an understanding, the technical solution of the present disclosure, in essence or the part that contributes to the related art, can be embodied in the form of a software product. The computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) and includes several instructions for causing an image display device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of the present disclosure.

[0170] The above is only a preferred embodiment of the present application and is not used to limit the protection scope of the present application.

Claims

1. A performance evaluation method, characterized in that: Applied to a distributed block storage system, the method comprises: When a cluster in the distributed block storage system is under load, obtaining a first data model used by the cluster and a performance value when the cluster uses the first data model to process the load; Matching the first data model with multiple data models to find a first static performance baseline value when the cluster uses the first data model to process the load from multiple static performance baseline values; the multiple static performance baseline values ​​are obtained when the cluster uses the multiple data models to process the load when the cluster is in an unloaded state; A dynamic performance baseline value when the cluster uses the first data model to perform load processing is determined according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

2. The method according to claim 1, characterized in that The method further comprises: When the cluster is unloaded, using the multiple data models to load-pressurize the cluster respectively; When the cluster is under load pressure, a dial test process is performed on the cluster to obtain multiple dial test results when the cluster uses the multiple data models to perform load processing; The multiple static performance baseline values ​​are determined according to the multiple dialing test results and preset performance baseline values.

3. The method according to claim 2, characterized in that The determining the multiple static performance baseline values ​​according to the multiple dialing test results and the preset performance baseline values ​​includes: Determine first delay information corresponding to a first dial test result among the multiple dial test results, and obtain second delay information in the preset performance baseline value; When the difference between the first delay information and the second delay information is less than a first threshold, the second data model corresponding to the first dialing result is searched from the multiple data models, and the pressure load value under the second data model is used as the static performance baseline value corresponding to the cluster when using the second data model for load processing.

4. The method according to claim 2, characterized in that: The method further comprises: Determine first delay information corresponding to a first dial test result among the multiple dial test results, and obtain second delay information in the preset performance baseline value; When the first delay information is greater than the second delay information and the difference between the first delay information and the second delay information is greater than a first threshold, load reduction is performed on the second data model corresponding to the first dialing test result; When the first delay information is less than or equal to the second delay information and a difference between the first delay information and the second delay information is greater than a first threshold, load pressure is applied to the second data model.

5. The method according to claim 1, characterized in that The determining, according to the performance value and the first static performance baseline value, a dynamic performance baseline value when the cluster uses the first data model to perform load processing includes: Performing a dial test on the cluster to obtain a second dial test result when the cluster uses the first data model to perform load processing; The first static performance baseline value is updated according to the second dial test result and the performance value to obtain a second static performance baseline value; The dynamic performance baseline value is determined based on the performance value and the second static performance baseline value.

6. The method according to claim 5, characterized in that The updating of the first static performance baseline value according to the second dialing test result and the performance value to obtain a second static performance baseline value includes: In the case where the performance value is greater than the first static performance baseline value, determining the third delay information corresponding to the second dial test result, and obtaining the second delay information in the preset performance baseline value; When a ratio between the third delay information and the second delay information is greater than a second threshold, using the first static performance baseline value as the second static performance baseline value; When the ratio of the third delay information to the second delay information is less than or equal to the second threshold, the first static performance baseline value is updated to the second static performance baseline value according to the performance value.

7. The method according to claim 5, characterized in that The updating of the first static performance baseline value according to the second dialing test result and the performance value to obtain a second static performance baseline value includes: In the case where the performance value is less than or equal to the first static performance baseline value, determining the third delay information corresponding to the second dial test result, and obtaining the second delay information in the preset performance baseline value; When the ratio between the third delay information and the second delay information is greater than a third threshold, outputting an alarm signal to check whether the cluster is abnormal; When no abnormality occurs in the cluster, the first static performance baseline value is updated to the second static performance baseline value according to the performance value.

8. The method according to claim 7, characterized in that After checking whether the cluster is abnormal, the method further includes: When an abnormality occurs in the cluster, fault processing is performed on the cluster.

9. The method according to claim 5, characterized in that The updating of the first static performance baseline value according to the second dialing test result and the performance value to obtain a second static performance baseline value includes: In the case where the performance value is less than or equal to the first static performance baseline value, determining the third delay information corresponding to the second dial test result, and obtaining the second delay information in the preset performance baseline value; When a ratio between the third delay information and the second delay information is less than or equal to a third threshold and greater than a second threshold, updating the first static performance baseline value to the second static performance baseline value according to the performance value; When the ratio of the third delay information to the second delay information is less than or equal to the second threshold, the first static performance baseline value is used as the second static performance baseline value.

10. A performance evaluation device, characterized in that: Applied to a distributed block storage system, the device comprises: an acquisition unit, configured to acquire, when a cluster in the distributed block storage system is under load, a first data model used by the cluster and a performance value when the cluster uses the first data model to process the load; A search unit, used for matching the first data model with a plurality of data models, so as to search, from a plurality of static performance baseline values, for a first static performance baseline value when the cluster uses the first data model to perform load processing; the plurality of static performance baseline values ​​are obtained when the cluster uses the plurality of data models to perform load processing when the cluster is in an unloaded state; An evaluation unit is used to determine a dynamic performance baseline value when the cluster uses the first data model to perform load processing according to the performance value and the first static performance baseline value, so as to evaluate the performance of the distributed block storage system.

11. A performance evaluation device, characterized in that: The method comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the steps of the method according to any one of claims 1 to 9 are implemented when the processor executes the program.

12. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images