DSS optimization configuration method and system based on influence rules between fitting performance indicators

Abstract

The application relates to the technical field of DSS optimization configuration, and particularly discloses a DSS optimization configuration method and system based on influence rules among fitting performance indexes, which comprises the following steps: collecting key performance index data and constructing priori knowledge rules; after pre-processing and index calculation are performed on the key performance index data, data of the performance index data is supplemented based on configurable items; the supplemented data is grouped and summarized to obtain a plurality of performance index data relationship mapping tables; the configurable items in each performance index data relationship mapping table are linearly fitted with the performance index data to obtain a first influence rule; whether the first influence rule conforms to the priori knowledge rules is judged; if not, the configurable items are subjected to multiple linear fitting with the performance index data to obtain a second influence rule; and the configuration of the distributed storage system is adjusted based on the first influence rule or the second influence rule. The method can help users to adjust the storage system configuration and realize load balancing.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of DSS optimization configuration, and particularly relates to a DSS optimization configuration method and system based on fitting performance index influence rules. BACKGROUND

[0002] The change of external application computing scenarios makes the internal structure and running environment of a distributed storage system (DSS) very complex, and the load characteristics change a lot, which makes the configuration of a load-balanced distributed storage system more complex, and the performance index evaluation of the distributed storage system (DSS) can provide guidance for the configuration of the distributed storage system; however, the current performance index research mostly stays in evaluating the correlation between performance indexes based on genetic algorithms and correlation mining algorithms; and since the traditional genetic algorithm principle and the correlation mining algorithm principle are step-by-step checking of random rules, the random rules do not have causality, and therefore are not suitable for mining the mutual influence relationship between DSS system indexes, and the traditional genetic algorithm principle can only check slowly and cannot achieve quantitative influence, and cannot provide optimization guidance for the configuration of the distributed storage system; and there is currently no effective method for regular checking and performance curve fitting of the mutual influence between performance indexes of the distributed storage system. SUMMARY

[0003] In view of the above problems, a first object of the present application is to provide a DSS optimization configuration method based on fitting performance index influence rules, which can effectively identify the mutual influence between indexes and fit the relevant rules, positively affect the accuracy of performance evaluation of actual application scenarios, help users adjust the storage system configuration, optimize the configuration scheme, optimize the performance of the distributed storage system, and achieve load balancing.

[0004] A second object of the present application is to provide a DSS optimization configuration system based on fitting performance index influence rules.

[0005] The first technical solution adopted by the present application is a DSS optimization configuration method based on fitting performance index influence rules, comprising the following steps:

[0006] S100: collecting key performance index data of a distributed storage system; and constructing priori knowledge rules; S200: pre-processing the key performance index data to obtain first performance index data; performing index calculation based on the first performance index data to obtain second performance index data; and performing data padding on the first performance index data and the second performance index data based on configurable items to obtain padded data; grouping and summarizing the padded data based on the priori knowledge rules to obtain a plurality of performance index data relationship mapping tables;

[0007] S300: Linearly fitting each configurable item in the performance indicator data relationship mapping table with each performance indicator data to obtain a first influence rule between the configurable item and each performance indicator;

[0008] S400: Determine whether the first influence rule between each configurable item and the performance indicator conforms to the priori knowledge rule, if it conforms, directly execute step S500; if it does not conform, perform multiple linear fitting on the configurable item and the performance indicator data to obtain a second influence rule between the configurable item and the performance indicator, and execute step S500;

[0009] S500: Optimizing and adjusting the configuration of the distributed storage system based on the first influence rule or the second influence rule.

[0010] Preferably, the key performance indicator data in step S100 includes one or more of the following data:

[0011] Total iops value under current configuration, total bandwidth under current configuration, user request feedback average time delay, total read times, cache read times, total write times and cache write times.

[0012] Preferably, the priori knowledge rule in step S100 includes: experience change trend curve of historical performance indicator inter-association rule, configuration factor and each performance indicator inter-association knowledge graph.

[0013] Preferably, the step S200 includes:

[0014] Cleaning, deduplication, outlier processing and consistency processing of the key performance indicator data to obtain first performance indicator data;

[0015] Calculating second performance indicator data based on the first performance indicator data through rule definition; the second performance indicator data includes but is not limited to cache hit rate and system total throughput.

[0016] Preferably, the configurable item in step S200 includes one or more of the following data:

[0017] Number of concurrent threads, mixed read-write ratio, file size, request access type, cache size, cache hit rate, system throughput, data consistency strategy type and number of redundant copies.

[0018] Preferably, the step S200 includes:

[0019] According to the correlation knowledge graph between the configuration factor in the priori knowledge rule and each performance index, the configurable item is taken as the independent variable, and the grouping principle of the performance index data under the current configurable item is taken as the dependent variable, the filling data is grouped, and a plurality of performance index data relationship mapping tables are formed.

[0020] Preferably, the step S300 comprises:

[0021] The configurable item in the performance index data relationship mapping table is taken as the independent variable, and each performance index data under the current configurable item is taken as the dependent variable, the configurable item and the performance index data in linear mapping relationship are linearly fitted based on the linear regression method, and the first influence rule between the configurable item and each performance index is obtained.

[0022] Preferably, the step S400 comprises:

[0023] All the first influence rules fitted in each performance index data relationship mapping table are compared with the priori knowledge rule, if the experience change trend curve of any first influence rule and the historical performance index correlation rule in the priori knowledge rule is the same or differs less than a set threshold, it is determined that the first influence rule conforms to the priori knowledge rule.

[0024] If the experience change trend curve of any first influence rule and the historical performance index correlation rule in the priori knowledge rule is contrary or differs more than or equal to a set threshold, it is determined that the first influence rule does not conform to the priori knowledge rule.

[0025] The second technical scheme adopted by the application is a DSS optimization configuration system based on fitted performance index influence rule, comprising a data acquisition module, a data induction and filling module, a performance index curve fitting module, a priori knowledge comparison module, a multiple linear fitting module and a configuration optimization module.

[0026] The data acquisition module is used for acquiring key performance index data of a distributed storage system.

[0027] The data induction and filling module is used for pre-processing the key performance index data to obtain first performance index data, performing index calculation based on the first performance index data to obtain second performance index data, and filling data based on the configurable item to obtain filling data, grouping and inducing the filling data based on the priori knowledge rule to obtain a plurality of performance index data relationship mapping tables.

[0028] The performance index curve fitting module is used for linearly fitting the configurable item and each performance index data in each performance index data relationship mapping table to obtain the first influence rule between the configurable item and each performance index.

[0029] The priori knowledge comparison module is configured to construct a priori knowledge rule, and determine whether a first influence rule between each configurable item and a performance index conforms to the priori knowledge rule, if yes, directly call the configuration change and scheme guidance module, if not, call the multiple linear fitting module to perform multiple linear fitting on the configurable item and the performance index data, obtain a second influence rule between the configurable item and the performance index, and then call the configuration change and scheme guidance module.

[0030] The configuration optimization module is configured to optimize and adjust the configuration of the distributed storage system based on the first influence rule or the second influence rule.

[0031] The above technical scheme has the following beneficial effects:

[0032] (1) The present application provides a DSS optimization configuration method based on fitting influence rules between performance indexes, which can effectively identify the mutual influence between indexes and fit the relevant rules, positively affect the accuracy of performance evaluation of actual application scenarios, help users adjust the configuration of the storage system, optimize the configuration scheme, optimize the performance of the distributed storage system, and achieve load balancing.

[0033] (2) The present application provides a DSS optimization configuration method based on fitting influence rules between performance indexes, which uses a fitting method of mutual influence rules between performance indexes of the distributed storage system, first maps the collected trace data into groups, draws scatter points and performs linear regression, then constructs a priori knowledge system, discovers the mutual influence relationship between indexes and performs multiple linear fitting, finally outputs a configuration change scheme guidance for users, realizes the discrimination of the correlation between indexes, completes the curve fitting of the rules, and supports the high-precision performance evaluation requirements of the distributed storage system.

[0034] (3) The present application can target the correlation mining of the complex performance index system of the distributed storage system, and through the linear fitting algorithm and the hierarchical regression fitting model, the mapping relationship between indexes is arranged and refined, the influence of the complexity of load characteristics and the change of configuration environment on the performance index of the distributed storage system is eliminated, and the evaluation accuracy is improved.

[0035] (4) The present application fits the mapping relationship between performance indexes through a linear regression curve fitting algorithm, and then compares with priori knowledge to determine whether there is a mutual influence relationship between key performance indexes, solving the problem that the relationship between complex indexes cannot be discriminated in the performance evaluation of the distributed storage system.

[0036] (5) The present application is aimed at an index data table with mutual influence relationship, based on the idea of hierarchical regression model, so that the user can add index items one by one, find out the conflict index items accurately by changing the configuration items (system running environment), and fit the mapping relationship between indexes in stages based on multiple tests, and finally give the rule expression; solve the pain point problem that the influence deviation between indexes cannot be quantified when testing the distributed storage system, and clearly express the change rule of the rule.

[0037] (6) The method architecture of the present application can give the user the quantitative change rule of the conflict index, assist the user to more accurately evaluate the performance index of the distributed storage system, and through the configurable module, the load change problem can be found in time, and the storage strategy can be adjusted in time. BRIEF DESCRIPTION OF DRAWINGS

[0038] Figure 1 A flowchart of a DSS optimization configuration method based on fitting performance index influence rules is provided for an embodiment of the present application;

[0039] Figure 2 A structure diagram of a DSS optimization configuration system based on fitting performance index influence rules is provided for an embodiment of the present application. DETAILED DESCRIPTION

[0040] The embodiments of the present application will be further described in detail below in combination with the drawings and examples. The detailed description of the following examples and the drawings are used to exemplarily illustrate the principles of the present application, but cannot be used to limit the scope of the present application, that is, the present application is not limited to the preferred embodiments described, and the scope of the present application is defined by the claims.

[0041] In the description of the present application, it should be noted that, unless otherwise specified, the meaning of "a plurality of" is two or more; the terms "first", "second", etc. are only used for description purposes, and cannot be understood as indicating or implying relative importance; the above-mentioned terms can be understood in the specific meaning in the present application by the person skilled in the art according to the specific circumstances.

[0042] Example 1

[0043] As Figure 1 shown, an embodiment of the present application provides a DSS optimization configuration method based on fitting performance index influence rules, comprising the following steps:

[0044] S100: collecting key performance index data of a distributed storage system; and constructing prior knowledge rules;

[0045] (1) Collecting key performance index data of a distributed storage system;

[0046] The Opentracing+Jaeger distributed full-link tracking tool is used to test and monitor the node state of the underlying storage device of the distributed storage system, and real-time collection and recording of the key performance indicator data (i.e., trace data) of the distributed storage system (DSS) are performed.

[0047] The present application mainly gives an output method of the correlation between the key performance indicators of the distributed storage system, mainly fitting the relationship between the key performance indicator data affecting the IOPS, latency and bandwidth of the storage system and the configurable items.

[0048] (2) constructing prior knowledge rules;

[0049] The prior knowledge rules include the experience change trend curve of the correlation between historical performance indicators, and the knowledge graph of the correlation between configuration factors and each performance indicator, i.e., the prior knowledge rules include the experience change trend rules between performance indicators and configuration factors, and the knowledge graph of the correlation between configuration factors and each performance indicator.

[0050] Further, in one embodiment, it further includes constructing the system stability and scalability change trend under typical configuration; based on the system stability and scalability change trend under the typical configuration, configuration guidance is performed under the typical scene load mode of four new computing modes of artificial intelligence, big data, graph computing and high-performance computing.

[0051] S200: preprocessing the collected key performance indicator data to obtain first performance indicator data; performing index calculation based on the first performance indicator data to obtain second performance indicator data; and performing data padding on the first performance indicator data and second performance indicator data based on configurable items to obtain padded data; grouping and summarizing the padded data based on the prior knowledge rules to obtain several performance indicator data relationship mapping tables;

[0052] The preprocessing of the key performance indicator data (i.e., trace data) includes cleaning, deduplication, outlier processing, consistency processing and other operations on the key performance indicator data to filter out valid key performance indicator data to obtain first performance indicator data; and constructing a basic performance indicator table based on the first performance indicator data.

[0053] The index calculation based on the first performance indicator data includes:

[0054] The first performance index data is calculated into second performance index data based on a rule definition (i.e. a second performance index calculation formula, such as a calculation formula of cache hit rate and system total throughput); the second performance index data includes but is not limited to cache hit rate, system total throughput;

[0055] The cache hit rate is used to analyze the influence law of cache capacity and cache hit rate on system IOPS and I / O time delay; the system total throughput is used to analyze the influence law between the system total throughput and I / O size (I / O size usually refers to block size, i.e. the data amount of single I / O, which is usually configurable at the user end).

[0056] The first performance index data and the second performance index data are data-complemented based on the configurable item, including:

[0057] The missing index data in the first performance index data and the second performance index data is data-complemented based on the configurable item (i.e. the complemented configurable item), to obtain complemented data; the complemented data includes the complemented configurable item, the first performance index data and the second performance index data.

[0058] The configurable item includes but is not limited to: user-end configurable item of hardware and load configuration item under current scene application; for the load configuration item under current scene application, the method provided by the application can be based on historical data to statistically, classify and configure the key parameter items of different request load read-write characteristics: r / w_rate (mixed read-write ratio), request access type (sequential I / O\random I / O), access file size, access frequency in special time sequence, etc., to simulate the load characteristics under scene application; the load configuration item is, for example, 100% random read, 100% random write, 100% sequential read, 100% sequential write, IOPS ratio minimum not less than 50% (half of read and half of write) and the like; that is, the configurable item includes but is not limited to: number of concurrent threads, r / w_rate (mixed read-write ratio), file size, request access type (sequential\random), cache size, cache hit rate, system throughput, data consistency strategy type, number of redundant copies and the like.

[0059] Further, it further includes generating a full-data performance query table based on the complemented data and storing it into a database, so that the user can real-time view the basic performance changes of the distributed storage system after changing the configuration.

[0060] The complemented data is grouped and summarized based on the priori knowledge rule, to obtain a plurality of performance index data relationship mapping tables, including:

[0061] According to the knowledge graph of the correlation between the configuration factors and the performance indicators in the priori knowledge rule, taking the configurable items (i.e., the configuration factors) as independent variables, and selecting the performance indicator data under the current configurable item from the supplemented data as the grouping principle of dependent variables, the supplemented data is grouped to form multiple performance indicator data relationship mapping tables; each performance indicator data relationship mapping table records the correlation between one or more configurable items and multiple performance indicator data; for example, a row in the performance indicator data relationship mapping table records the configurable items, and each column records different performance indicator data.

[0062] S300: Linearly fitting the configurable items and the performance indicator data in each performance indicator data relationship mapping table to obtain a first influence rule between the configurable items and the performance indicators;

[0063] Linearly fitting the configurable items and the performance indicator data in each performance indicator data relationship mapping table includes:

[0064] Taking the configurable items in the performance indicator data relationship mapping table as independent variables and taking the performance indicator data under the current configurable item as dependent variables, the relationship between the configurable items and the performance indicator data is preliminarily judged by drawing a scatter plot, the configurable items and the performance indicator data that present a linear mapping relationship are linearly fitted based on a linear regression method to obtain a first influence rule between the configurable items and the performance indicators, and the linear expression of the first influence rule is:

[0065] h θ =X T θ=x0θ0+x1θ1

[0066] In the formula, hθ is a performance indicator (i.e., a dependent variable); X T is a configurable item (an independent variable); θ is the value of a weight matrix when a loss function reaches a minimum value, i.e., a fitting gradient of a linear expression; x0 and x1 are different configurable items; θ0 and θ1 are fitting gradients between the performance indicators and different configurable items.

[0067] S400: Judging whether the first influence rule between each configurable item and a performance indicator conforms to the priori knowledge rule, if yes, directly performing step S500; if not, performing multiple linear fitting on the configurable item and the performance indicator data to obtain a second influence rule between the configurable item and the performance indicator, and performing step S500;

[0068] Judging whether the first influence rule between each configurable item and a performance indicator in each performance indicator data relationship mapping table conforms to the priori knowledge rule includes:

[0069] The first influence rule between all configurable items and performance indicators fitted in each performance indicator data relationship mapping table is compared with the priori knowledge rule; if the first influence rule between any configurable item and performance indicator is the same as or differs less than a set threshold (the set threshold is set according to experience) from the experience change trend curve of the historical performance indicator inter-association rule in the priori knowledge rule, it is determined that the first influence rule between the configurable item and the performance indicator conforms to the priori knowledge rule (namely, the mapping relationship between the configurable item and each performance indicator data in the performance indicator data relationship mapping table is normal);

[0070] If the first influence rule between any configurable item and performance indicator is contrary to or differs more than / equal to a set threshold (the set threshold is set according to experience) from the experience change trend curve of the historical performance indicator inter-association rule in the priori knowledge rule, it is determined that the first influence rule between the configurable item and the performance indicator does not conform to the priori knowledge rule.

[0071] The configurable item and performance indicator data corresponding to the priori knowledge rule that does not conform are re-associated and configured for analysis and new rule change equation fitting, namely, multiple linear fitting.

[0072] S500: Based on the first influence rule or the second influence rule between the configurable item and each performance indicator, a user is guided to optimize and adjust the configuration of the distributed storage system.

[0073] The user can change the configurable item according to the provided configuration interface; the configurable item includes but is not limited to: the number of concurrent threads, r / w_rate (mixed read-write ratio), file size, request access type (sequential \ random), cache size, cache hit rate, system throughput, data consistency policy type, redundancy copy number, etc.

[0074] When the user changes the configuration, the configurable item configured by the user is first decomposed, the performance indicator associated with the configurable item is confirmed, and then the first influence rule or the second influence rule between the configurable item and each performance indicator is combined to feed back to the user the associated performance indicator and performance indicator change law caused after the configurable item is changed, help the user to adjust the storage system configuration, optimize the configuration scheme, optimize the performance of the distributed storage system, and realize load balancing; after the user changes the configuration, the related indicators of the running distributed storage system can be queried.

[0075] Embodiment two

[0076] As shown in Figure 2 An embodiment of the present application provides a DSS optimization configuration system based on fitted performance indicator influence rules, which comprises a data acquisition module, a data induction and supplement module, a performance indicator curve fitting module, a priori knowledge comparison module, a multiple linear fitting module and a configuration optimization module.

[0077] The data collection module is configured to collect key performance indicator data of the distributed storage system.

[0078] The data induction and completion module provides an automatic calculation function, is configured to pre-process the key performance indicator data to obtain first performance indicator data, perform indicator calculation based on the first performance indicator data to obtain second performance indicator data, perform data completion on the first performance indicator data and the second performance indicator data based on configurable items to obtain completion data, group and induce the completion data based on the priori knowledge rules to obtain a plurality of performance indicator data relationship mapping tables.

[0079] The performance indicator curve fitting module is configured to perform linear fitting on the configurable items and the performance indicator data in each performance indicator data relationship mapping table to obtain a first influence rule between the configurable items and the performance indicators.

[0080] The priori knowledge comparison module is configured to construct a priori knowledge rule, and determine whether the first influence rule between each configurable item and the performance indicator conforms to the priori knowledge rule. If yes, the configuration change and scheme guidance module is directly invoked. If not, the multi-element linear fitting module is invoked to perform multi-element linear fitting on the configurable item and the performance indicator data to obtain a second influence rule between the configurable item and the performance indicator, and then the configuration change and scheme guidance module is invoked.

[0081] The multi-element linear fitting module is configured to perform multi-element linear fitting on the configurable item and the performance indicator data to obtain a second influence rule between the configurable item and the performance indicator.

[0082] The configuration optimization module (i.e. the configuration change and scheme guidance module) is configured to guide a user to optimize and adjust the configuration of the distributed storage system based on the first influence rule or the second influence rule between the configurable item and the performance indicators.

[0083] The DSS optimization configuration system based on the fitted influence rule between performance indicators further comprises a client system configuration module and a performance indicator query module.

[0084] The client system configuration module is configured to provide a configuration interface for the user, and the performance indicator query module is configured to support the user to query related performance indicators of the distributed storage system.

[0085] Those skilled in the art can appreciate that the units and algorithm steps of each example described in combination with the embodiments disclosed in the present application can be realized in electronic hardware or in a combination of computer software and electronic hardware. Whether the functions are realized in hardware or software depends on the specific application and design constraints of the technical solution. A person skilled in the art can use different methods to realize the described functions for each specific application, but such implementation should not be considered beyond the scope of the present application.

[0086] In the embodiments provided in the present application, it should be understood that the disclosed apparatus and method can be implemented in other ways. For example, the apparatus embodiments described above are only schematic. The division of the units is only a logical function division. There can be another division manner in actual implementation, for example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual coupling or direct coupling or communication connection can be indirect coupling or communication connection through some interface, device or unit, and can be electrical, mechanical or in other forms.

[0087] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e. they can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0088] In addition, each functional unit in each embodiment of the present application can be integrated into a processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0089] If the functions are realized in the form of software function units and sold or used as independent products, they can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the present application or the parts of the technical solutions that essentially contribute to the prior art can be embodied in the form of software products. The computer software product is stored in a storage medium and includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a ROM, a RAM, a magnetic disk or an optical disk, and various storage media that can store program codes.

[0090] The above merely illustrates the specific embodiments of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of the changes or replacements within the technical range disclosed by the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A DSS optimization configuration method based on the influence rules between fitting performance indicators, characterized in that, Includes the following steps: S100: Collects key performance indicator data of distributed storage systems; And the construction of prior knowledge rules; S200: Preprocess the key performance indicator data to obtain first performance indicator data; and perform indicator calculation based on the first performance indicator data to obtain second performance indicator data; And based on configurable items, the first performance index data and the second performance index data are supplemented to obtain supplemented data; Based on the prior knowledge rules, the completed data is grouped and summarized to obtain several performance index data relationship mapping tables. S300: Linearly fit the configurable items in each performance index data relationship mapping table with each performance index data to obtain the first influence rule between the configurable items and each performance index. S400: Determine whether the first influence rule between each configurable item and the performance indicator conforms to the prior knowledge rule. If it does, proceed directly to step S500. If it does not, perform multivariate linear fitting on the configurable item and the performance indicator data to obtain the second influence rule between the configurable item and the performance indicator, and then proceed to step S500. S500: Optimize and adjust the configuration of the distributed storage system based on the first or second influence rule.

2. The DSS optimization configuration method according to claim 1, characterized in that, The key performance indicator data in step S100 includes one or more of the following: The current configuration includes total IOPS, total bandwidth, average latency of user request responses, total number of reads, number of cache reads, total number of writes, and number of cache writes.

3. The DSS optimization configuration method according to claim 1, characterized in that, The prior knowledge rules in step S100 include: empirical trend curves of association rules between historical performance indicators, and knowledge graphs of the association between configuration factors and various performance indicators.

4. The DSS optimization configuration method according to claim 1, characterized in that, Step S200 includes: The key performance indicator data is cleaned, deduplicated, outlier-handled, and consistency-handled to obtain the first performance indicator data. The second performance indicator data is calculated based on the first performance indicator data through rule definition; the second performance indicator data includes, but is not limited to, cache hit rate and total system throughput.

5. The DSS optimization configuration method according to claim 1, characterized in that, The configurable items in step S200 include one or more of the following data: Concurrent threads, mixed read / write ratio, file size, request access type, cache size, cache hit rate, system throughput, data consistency strategy type, and number of redundant replicas.

6. The DSS optimization configuration method according to claim 1, characterized in that, Step S200 includes: Based on the knowledge graph of the relationship between configuration factors and various performance indicators in the prior knowledge rules, the supplementary data is grouped according to the grouping principle of taking the configurable item as the independent variable and the performance indicator data under the current configurable item in the supplementary data as the dependent variable, to form multiple performance indicator data relationship mapping tables.

7. The DSS optimization configuration method according to claim 1, characterized in that, Step S300 includes: Using the configurable items in the performance index data relationship mapping table as independent variables and the performance index data under the current configurable item as dependent variables, the configurable items that show a linear mapping relationship and the performance index data are linearly fitted using the linear regression method to obtain the first influence rule between the configurable items and each performance index.

8. The DSS optimization configuration method according to claim 1, characterized in that, Step S400 includes: All first influence rules fitted in each performance index data relationship mapping table are compared with the prior knowledge rules; if any first influence rule is the same as or differs from the empirical change trend curve of the historical performance index association rule in the prior knowledge rules by less than a set threshold, then the first influence rule is determined to conform to the prior knowledge rules. If any first influence rule contradicts or differs from the empirical trend curve of the correlation rule between historical performance indicators in the prior knowledge rule, or if the difference is greater than or equal to a set threshold, then the first influence rule is determined to be inconsistent with the prior knowledge rule.

9. A DSS optimization configuration system based on the influence rules between fitting performance indices, characterized in that, It includes a data acquisition module, a data summarization and completion module, a performance index curve fitting module, a prior knowledge comparison module, a multiple linear fitting module, and a configuration optimization module. The data acquisition module is used to collect key performance indicator data of the distributed storage system. The data summarization and completion module is used to preprocess the key performance indicator data to obtain first performance indicator data; and to perform indicator calculation based on the first performance indicator data to obtain second performance indicator data; and to complete the first performance indicator data and second performance indicator data based on configurable items to obtain completed data. Based on prior knowledge rules, the completed data is grouped and summarized to obtain several performance index data relationship mapping tables. The performance index curve fitting module is used to perform linear fitting between the configurable items in each performance index data relationship mapping table and each performance index data to obtain the first influence rule between the configurable items and each performance index. The prior knowledge comparison module is used to construct prior knowledge rules; and to determine whether the first influence rule between each configurable item and the performance indicator conforms to the prior knowledge rules. If it does, the configuration optimization module is directly called; if it does not, the multiple linear fitting module is called to perform multiple linear fitting between the configurable item and the performance indicator data to obtain the second influence rule between the configurable item and the performance indicator, and then the configuration optimization module is called. The configuration optimization module is used to optimize and adjust the configuration of the distributed storage system based on the first influence rule or the second influence rule.

Images