System performance inspection method and device, storage medium and computer equipment

Abstract

According to the system performance inspection method and device, the storage medium and the computer equipment provided by the invention, firstly, the performance index data is preliminarily analyzed through the first model, original fixed rule judgment is replaced, and preliminary understanding of a performance index trend is realized. Then, a matched historical optimal case and an analysis framework are retrieved in the knowledge base according to the preliminary analysis trend, and empirical contexts and judgment standards are injected for the current decision; and finally, inputting the performance index data, the optimal case and the analysis framework into a second model, and simulating a decision-making process of performing comprehensive research and judgment by an expert in combination with experience, a method and real-time data through the second model. And a highly accurate inspection report is obtained. In the process, in combination with deep analysis of the deep learning model and the knowledge base, the ability of understanding and judging performance index fluctuation is improved, so that the misjudgment probability is reduced, and the accuracy and reliability of performance inspection are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of data processing, and particularly relates to a system performance inspection method and device, a storage medium and computer equipment. BACKGROUND

[0002] In the field of software system operation and maintenance, performance inspection is a key link to ensure stable operation of the system. The existing technology mainly relies on an automatic inspection system to collect multiple performance indicators such as CPU usage and interface response time at regular intervals, and to determine whether there is a continuous deterioration trend of the indicators within a specific time window based on preset algorithm rules or statistical models, so as to realize automatic discovery and alarm of problems. This method aims to replace inefficient manual inspection and improve the timeliness of operation and maintenance response.

[0003] However, since performance data is affected by multiple factors and fluctuation scenarios are complex and diverse, the existing scheme is based on fixed logic and lacks deep understanding of the overall context and trend evolution of the data. This leads to false positives when the system faces common scenarios such as normal fluctuations in indicators and rapid recovery after occasional spikes, thereby resulting in low accuracy of performance inspection. SUMMARY

[0004] The present application aims to at least solve one of the above technical defects, particularly the technical defect that the existing technology is based on fixed logic and lacks deep understanding of the overall context and trend evolution of the data. This leads to false positives when the system faces common scenarios such as normal fluctuations in indicators and rapid recovery after occasional spikes, thereby resulting in low accuracy of performance inspection.

[0005] In a first aspect, the present application provides a system performance inspection method, which comprises:

[0006] obtaining performance indicator data of a target system;

[0007] inputting the performance indicator data into a preset first model to obtain a preliminary analysis trend;

[0008] determining a preset knowledge base and searching the knowledge base using the preliminary analysis trend to obtain a best case and an analysis framework matched with the preliminary analysis trend;

[0009] inputting the performance indicator data, the best case and the analysis framework into a preset second model to obtain an inspection report of the target system.

[0010] In one of the embodiments, the inputting the performance indicator data into a preset first model to obtain a preliminary analysis trend comprises:

[0011] obtaining a preset prompt word;

[0012] inputting the prompt word and the performance index data into a preset first model;

[0013] performing trend analysis on the performance index data according to the prompt word by the first model to identify change patterns and abnormal fluctuations of performance indexes, and generating a preliminary analysis trend according to the change patterns and the abnormal fluctuations.

[0014] In one embodiment, the updating process of the knowledge base comprises:

[0015] obtaining historical inspection reports;

[0016] performing structured processing on the historical inspection reports to extract performance data summaries;

[0017] determining a preset knowledge base, and writing the historical inspection reports into a category corresponding to the index type of the performance data summaries in the knowledge base to update the knowledge base.

[0018] In one embodiment, the searching in the knowledge base by using the preliminary analysis trend to obtain an optimal case and an analysis framework matching the preliminary analysis trend comprises:

[0019] analyzing the preliminary analysis trend to determine a search keyword;

[0020] performing similarity matching in the knowledge base based on the search keyword, and determining an inspection case in the knowledge base having the highest similarity to the search keyword as the optimal case;

[0021] extracting analysis steps and judgment criteria of the optimal case, and generating an analysis framework according to the analysis steps and the judgment criteria.

[0022] In one embodiment, the inputting of the performance index data, the optimal case and the analysis framework into a preset second model to obtain an inspection report of the target system comprises:

[0023] inputting the performance index data into a preset second model, and providing the optimal case and the analysis framework as context reference information to the second model;

[0024] performing deep analysis on the performance index data by the second model and using the context reference information to generate an inspection report containing a performance data summary, detailed analysis content and an inspection conclusion.

[0025] In one of the embodiments, the performance index data is deeply analyzed by using the context reference information to generate a patrol report including performance data summary, detailed analysis content and patrol conclusion, which includes:

[0026] determining a key judgment rule from the context reference information;

[0027] analyzing the performance index data according to the key judgment rule to obtain performance data summary, detailed analysis content and patrol conclusion, and a patrol report of the target system;

[0028] wherein the key judgment rule includes:

[0029] if the performance index data keeps below a preset threshold error rate in a monitoring period of recent five or more consecutive days, it is determined that the target system has no degradation trend.

[0030] In one of the embodiments, the method further includes:

[0031] analyzing the patrol report of the target system, and if the analyzed patrol conclusion is system degradation, sending the patrol report to a corresponding terminal.

[0032] In a second aspect, the application provides a system performance patrol device, which includes:

[0033] a data acquisition module for acquiring performance index data of a target system;

[0034] a trend analysis module for inputting the performance index data into a preset first model to obtain a preliminary analysis trend;

[0035] a knowledge retrieval module for determining a preset knowledge base and retrieving in the knowledge base by using the preliminary analysis trend to obtain a best case and an analysis framework matching the preliminary analysis trend;

[0036] a report generation module for inputting the performance index data, the best case and the analysis framework into a preset second model to obtain a patrol report of the target system.

[0037] In a third aspect, the application provides a storage medium having computer readable instructions stored therein, which, when executed by one or more processors, cause the one or more processors to perform the steps of the system performance patrol method according to any one of the above embodiments.

[0038] In a fourth aspect, the application provides a computer device including one or more processors and a memory.

[0039] The memory stores computer readable instructions, and the one or more processors execute the computer readable instructions to perform the steps of the system performance inspection method of any of the above embodiments.

[0040] From the above technical solutions, the embodiments of the present application have the following advantages:

[0041] The system performance inspection method, device, storage medium and computer equipment provided by the present application first perform preliminary analysis on the performance index data through the first model, replacing the original fixed rule judgment, and realizing preliminary understanding of the performance index trend. Then, the matched historical best case and analysis framework are searched in the knowledge base according to the preliminary analysis trend, and experience context and judgment standard are injected for the current decision. Finally, the performance index data, the best case and the analysis framework are input into the second model, and the decision-making process of experts combining experience, method and real-time data for comprehensive research and judgment is simulated through the second model. A highly accurate inspection report is obtained. In the above process, the deep analysis of the deep learning model and the knowledge base is combined to improve the understanding and judgment ability of the performance index fluctuation, thereby reducing the misjudgment probability and improving the accuracy and reliability of the performance inspection. BRIEF DESCRIPTION OF DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiments or the prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor.

[0043] Figure 1 A flowchart of a system performance inspection method provided by an embodiment of the present application;

[0044] Figure 2 A structural diagram of a system performance inspection device provided by an embodiment of the present application;

[0045] Figure 3 An internal structure diagram of a computer equipment provided by an embodiment of the present application. DETAILED DESCRIPTION

[0046] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0047] In one of the embodiments, the present application provides a system performance inspection method, and the following embodiments are described with the method applied to a server. It can be understood that the system performance inspection method can be performed on a single server or a server cluster composed of multiple servers, and the present application does not make specific limitations thereto.

[0048] As shown in Figure 1 The present application provides a system performance inspection method, which comprises:

[0049] S101: Obtain performance index data of a target system.

[0050] The target system refers to a software application system, including but not limited to a website application, a terminal application, etc. The performance index data refers to various quantitative parameters for evaluating the performance of the target system within a period of time, such as CPU usage, abnormal error rate, etc.

[0051] In this step, when the performance of a certain system needs to be inspected, the performance index data can be collected through the performance monitoring tool built in the system or the third-party monitoring tool. For example, in the Java environment, JMX (Java Management Extensions) can be used to obtain the performance index of JVM, or professional APM (Application Performance Management) tools such as New Relic, Dynatrace, etc. can be used to collect more extensive performance data. And the collected performance index data is transmitted to the server to obtain the performance index data of the target system.

[0052] Further, the performance inspection task can be set as a timing period task, which automatically triggers the performance inspection of the target system after a certain time interval. In addition, when determining the specific index of the performance index data, the function type and use scenario of the target system can be considered. For example, an online e-commerce platform may pay special attention to the interface response time of transaction processing and the throughput of the system to ensure that users can obtain a good shopping experience during peak hours. While a big data processing system may pay more attention to the use of CPU and memory, as well as the performance of disk I / O, to optimize the efficiency of data processing.

[0053] S102: Input the performance index data into a preset first model to obtain a preliminary analysis trend.

[0054] The first model is used to preliminarily analyze the performance index data to determine the change trend of the performance index data within a period of time.

[0055] In this step, after obtaining the performance indicator data, the performance indicator data can be preprocessed to ensure that the data input into the first model is accurate and available. Then, the preprocessed performance indicator data is input into the first model, which can understand and analyze the performance indicator data semantically, and output the trend of the performance indicator data in the time range it covers, i.e., the preliminary analysis trend.

[0056] Specifically, the first model in the present application can adopt a large language model that has been fine-tuned. The large language model is based on a large-scale language model that has been specifically optimized and adjusted to adapt to specific tasks or application scenarios. It can be understood that a large-scale language model generally refers to a model that has been trained on a large amount of text data, which can understand and generate natural language and has a wide range of language understanding and generation capabilities. When such a model is fine-tuned to adapt to specific tasks, such as performance indicator data analysis, it can more accurately understand and process related data.

[0057] S103: Determine the preset knowledge base and use the preliminary analysis trend to search in the knowledge base to obtain the best case and analysis framework matching the preliminary analysis trend.

[0058] Among them, the knowledge base refers to a structured information collection that is pre-constructed and continuously maintained, which stores historical inspection cases including performance data summaries, detailed analysis contents and inspection conclusions. The best case refers to a historical inspection case with the highest similarity or reference value to the current preliminary analysis trend of the system. The analysis framework refers to a pattern of structured logic, key consideration dimensions and judgment criteria for guiding performance analysis.

[0059] In this step, after obtaining the preliminary analysis trend of the performance indicator data using the first model, relevant search can be performed in the preset knowledge base to query the best case matching the preliminary analysis trend and obtain the analysis framework of the best case to determine the best case and analysis framework matching the preliminary analysis trend.

[0060] Further, when performing relevant search in the knowledge base, the best case can be determined by calculating the similarity of the preliminary analysis trend and each case in the knowledge base. Specifically, when calculating the similarity, vector similarity, keyword matching, etc. can be considered.

[0061] Exemplarily, in the inspection of a containerized payment system, assuming that the preliminary analysis trend given by the first model is that the JVM memory usage of a certain core service presents a stepwise increase in the past week, and the frequency of Full GC is also increasing. The server will first determine to call the special knowledge base on the topic of "Java application memory and GC". Then, using features such as "stepwise increase" and "increasing frequency of Full GC", the inspection cases with similar data patterns in history are retrieved in the knowledge base. Thus, the existing solutions or adjustment strategies are quickly applied.

[0062] S104: input the performance indicator data, the best case and the analysis framework into the preset second model to obtain an inspection report of the target system.

[0063] The second model is used to perform deep analysis on the performance indicator data according to the best case and the analysis framework, and output an inspection report of the target system according to the analysis result. The inspection report at least contains a summary description of the performance data, detailed analysis content of the trend and potential problems, and a final inspection conclusion about whether there is a performance degradation trend.

[0064] In this step, the collected performance indicator data, the best case retrieved from the knowledge base and the analysis framework are input into the preset second model. The second model can comprehensively analyze the performance indicator data, determine the performance data summary, detailed analysis content and inspection conclusion, and thus generate the inspection report.

[0065] Specifically, the second model in the present application can adopt a fine-tuned large language model, which is optimized and adjusted on the basis of the original large-scale language model to adapt to specific tasks or application scenarios.

[0066] In the above embodiment, the performance indicator data is first analyzed by the first model, replacing the original fixed rule judgment, and realizing the preliminary understanding of the performance indicator trend. Then, the matching historical best case and analysis framework are retrieved in the knowledge base according to the preliminary analysis trend, injecting the experiential context and judgment standard for the current decision. Finally, the performance indicator data, the best case and the analysis framework are input into the second model, which simulates the decision-making process of experts combining experience, methods and real-time data for comprehensive research and judgment. A highly accurate inspection report is obtained. In the above process, the deep learning model and the knowledge base are combined for deep analysis, which improves the understanding and judgment ability of the performance indicator fluctuation, thereby reducing the misjudgment probability and improving the accuracy and reliability of the performance inspection.

[0067] In one of the embodiments, the performance indicator data is input into the preset first model to obtain a preliminary analysis trend, including:

[0068] S1: Obtain a preset prompt word.

[0069] S2: Input the prompt word and performance indicator data into a preset first model.

[0070] S3: Through the first model, perform trend analysis on the performance indicator data according to the prompt word to identify the change pattern and abnormal fluctuation of the performance indicator, and generate a preliminary analysis trend according to the change pattern and abnormal fluctuation.

[0071] The prompt word is a structured text instruction used to guide and constrain the thinking process, analysis role and output format of the first model. The change pattern refers to the form or trajectory of the performance indicator data over time, which has a certain regularity or typicality. The abnormal fluctuation refers to the sudden change of the performance indicator data at a specific time point, which significantly deviates from its historical regular range or expected change pattern.

[0072] In this embodiment, the prompt word related to performance analysis can be determined first, which should be able to accurately reflect the performance aspects that need to be concerned and the specific analysis tasks. Then, these prompt words are input into the preset first model together with the performance indicator data. The first model will perform trend analysis on the performance indicator data according to these prompt words, identify the change pattern and abnormal fluctuation of the performance indicator, and finally generate a preliminary analysis trend according to the identified change pattern and abnormal fluctuation. In the above process, by identifying the change pattern and abnormal fluctuation, potential performance problems can be discovered earlier, thereby improving the accuracy and reliability of performance inspection.

[0073] Specifically, when fine-tuning the first model of the present application, performance data samples and their preliminary analysis trends can be obtained, and based on these data, the large language model is fine-tuned, and the performance of the model is continuously evaluated during the fine-tuning process to adjust the large language model according to the evaluation results until the fine-tuning is completed to obtain the first model.

[0074] In one example, the prompt word can be represented as: "This is the abnormal error rate of 180 days, the unit is percentage, you are a performance test expert, if the error rate of a certain date is 0, it means that no error exception occurred on that day, please draw a curve according to the data, analyze whether there is a degradation trend in this system, the abnormal error rate is continuously rising, which is not normal, occasional sudden increase and then decrease is normal, but the overall trend curve is rising, which is not normal."

[0075] In one of the embodiments, the updating process of the knowledge base includes:

[0076] S1: Obtain historical inspection reports.

[0077] S2: Structuring the historical inspection report to extract a performance data summary.

[0078] S3: Determining a preset knowledge base, and writing the historical inspection report into a category corresponding to the index type of the performance data summary in the knowledge base to update the knowledge base.

[0079] The performance data summary refers to a condensed description of performance index data.

[0080] In this embodiment, the natural language processing technology or the parsing method based on the predefined template can be used to automatically identify and extract the core paragraphs or fields describing the data facts in the historical inspection report. The dispersed information obtained after the above structured processing is further summarized, integrated and standardized to form the performance data summary. Then, the category corresponding to the index type of the performance data summary is queried in the knowledge base. For example, if the summary mainly describes the error rate index, it may correspond to the error rate degradation category; if it mainly describes the CPU usage, it may correspond to the resource usage category. Then the historical inspection report is written into the category to update the knowledge base, that is, the complete report document (or together with its performance data summary) is stored as a new case entry in the determined category directory of the knowledge base, thereby expanding the case reserve of the knowledge base. Moreover, by updating the knowledge base, more accurate reference basis can be provided for future performance analysis and decision-making.

[0081] In one of the embodiments, the preliminary analysis trend is used to search in the knowledge base to obtain the best case and analysis framework matching the preliminary analysis trend, including:

[0082] S1: Analyzing the preliminary analysis trend to determine the search keywords.

[0083] S2: Performing similarity matching in the knowledge base based on the search keywords, and determining the inspection case in the knowledge base with the highest similarity to the search keywords as the best case.

[0084] S3: Extracting the analysis steps and judgment criteria of the best case, and generating an analysis framework according to the analysis steps and judgment criteria.

[0085] The search keywords refer to the words used to identify and locate specific information when performing information retrieval. The judgment criteria refer to the judgment conditions related to the inspection conclusion.

[0086] In this embodiment, the preliminary analysis trend is analyzed to determine the object and performance indicators of this performance inspection, and these information is used as the search keyword. When the search keyword is determined, similarity matching is performed in the knowledge base to find the most relevant inspection cases to the search keyword. Then, the inspection cases in the knowledge base with the highest similarity to the search keyword are determined as the best cases, and the analysis steps and judgment criteria of these cases are extracted. Through the analysis steps and judgment criteria, the accuracy and efficiency of performance analysis are improved, and the continuous accumulation and optimization of knowledge are also promoted, thereby ensuring the long-term stable operation of the system.

[0087] Specifically, when calculating the similarity of the inspection cases in the knowledge base and the search keyword, the semantic vectors of the two can be extracted respectively, and then the similarity between the inspection cases and the search keyword is determined according to the vector similarity calculation.

[0088] In one embodiment, the performance indicator data, the best case and the analysis framework are input into a preset second model to obtain an inspection report of the target system, including:

[0089] S1: input the performance indicator data into the preset second model, and provide the best case and the analysis framework as context reference information to the second model.

[0090] S2: perform deep analysis on the performance indicator data by the second model and using the context reference information, and generate an inspection report containing performance data summary, detailed analysis content and inspection conclusion.

[0091] The inspection conclusion includes degradation trend and no degradation trend. The context reference information refers to the background information or auxiliary information related to the data provided when performing data analysis. The detailed analysis content refers to the detailed results obtained after in-depth analysis of the performance indicator data, including performance trend, problem diagnosis, impact assessment, etc.

[0092] In this embodiment, the best case and the analysis framework are provided as context reference information to the second model, which will help the model to more accurately understand and analyze the performance indicator data. According to the guidance of the context reference information, the performance indicator data is analyzed in depth, and the performance data summary, the detailed analysis content and the inspection conclusion are generated in turn, so as to form the inspection report according to the performance data summary, the detailed analysis content and the inspection conclusion. In the above process, the case and the framework are input as context reference information together with the original data, which essentially injects the verified professional thinking mode and decision path into the analysis process of the model, guiding it to reason in a direction defined by structured professional knowledge, which greatly improves the depth and accuracy of analyzing complex performance problems.

[0093] In one of the embodiments, the performance indicator data is deeply analyzed by using the context reference information to generate a patrol report containing performance data summary, detailed analysis content and patrol conclusion, including:

[0094] S1: Determine the key judgment rule from the context reference information.

[0095] S2: Analyze the performance indicator data according to the key judgment rule to obtain the performance data summary, detailed analysis content and patrol conclusion, and the patrol report of the target system.

[0096] Among them, the key judgment rule refers to a series of pre-defined standards or conditions used to evaluate whether the system performance is deteriorating or meeting expectations when analyzing performance data. It includes: if the performance indicator data remains below the preset threshold error rate for more than five consecutive days or more monitoring periods, it is determined that the target system has no deterioration trend.

[0097] In this embodiment, the judgment standard is obtained from the context reference information as the key judgment rule, and then the performance indicator data is analyzed according to the key judgment rule to obtain the performance data summary, detailed analysis content and patrol conclusion, and the patrol report of the target system.

[0098] Specifically, the key judgment rule also includes: abnormal error rate is always rising is an abnormal phenomenon, and abnormal error rate occasionally surges and then falls is also a normal phenomenon.

[0099] In one example, a patrol report can be represented as:

[0100] (1) Performance data summary

[0101] Data period: 2025-04-17 to 2025-10-05 (about 180 days)

[0102] Error rate range: 0.039% to 40.198%

[0103] Key observations:

[0104] First stage: healthy period (mid-April to early May): the system is healthy, the error rate is basically below 1%, and the occasional single-day peak (such as April 23) can quickly recover, showing good fault tolerance and self-healing ability.

[0105] Second stage: serious deterioration period (mid-May to August): the system enters an unstable state, the error rate jumps from May 15 and fluctuates at a high level of 20%-40% for a long time, which is a clear deterioration process lasting about 3.5 months.

[0106] Third stage: Recovery (end of September - early October): The error rate dropped to a healthy level near 0% on September 30 and remained stable.

[0107] Overall trend: The data presents a complete "healthy → deteriorating → recovery" life cycle. Although there was a serious deterioration in the middle period, the end-of-period data shows that the deterioration problem has been successfully resolved, and the system performance has recovered and may even exceed the initial level. Therefore, from the final state of the entire time series, the deterioration trend did not continue.

[0108] (2) Detailed analysis content

[0109] Reinterpretation of end-of-period data:

[0110] The error rate dropped from a long-term high of over 20% to below 0.1%. This change is not a natural fluctuation, but a result of effective and fundamental repairs, such as deploying critical patches, fixing core code defects, expanding key resources, or resolving external dependency issues.

[0111] In performance engineering, a problem that has been repaired through intervention is considered to have ended its deterioration trend. When evaluating trends, we focus more on the current and final states. Since the root cause of the problem has been resolved, the upward trend caused by the problem has ended.

[0112] Stage definition of trend:

[0113] Local trend: From mid-May to August, the system had a significant deterioration trend, which cannot be denied.

[0114] Global trend: Observing the entire span from the starting point (April 17) to the end point (October 5), the system's error rate started from a low point, experienced a serious problem period, and finally returned to a low point. The starting and ending points are both healthy, so the global trend of the entire period is not continuously rising.

[0115] Key basis for conclusion:

[0116] The "deterioration trend" in performance testing usually refers to a state that is not resolved, continuously deteriorating, or persistently high. The high error rate problem in this system did not persist until the end of the data period, but was completely resolved at the end of the period.

[0117] Therefore, the system experienced a deterioration period, but the problem has been repaired and there is no deterioration trend at present.

[0118] (3) Final conclusion

[0119] Inspection conclusion: No deterioration trend.

[0120] Conclusion explanation: Although the system experienced a serious performance degradation for several months in the middle term, the error rate has recovered to a healthy level at the end of the monitoring period through effective intervention measures. This indicates that the root cause leading to the degradation has been eliminated, so the system does not have a persistent degradation trend from the final result of the entire time series. It is recommended to continue monitoring to verify the long-term effectiveness of the repair.

[0121] In one of the embodiments, the system performance inspection method further comprises:

[0122] analyzing the inspection report of the target system, and if the analysis result is system degradation, sending the inspection report to the corresponding terminal.

[0123] In this embodiment, when the analysis result is system degradation, the corresponding responsible persons can be determined according to the target system, and the inspection report and the analysis result are sent to the terminals corresponding to the responsible persons, so as to remind the responsible persons to take corresponding measures.

[0124] It should be understood that although each step in the flowchart involved in each embodiment as described above is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified herein, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in the flowchart involved in each embodiment as described above can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or steps or stages in other steps.

[0125] The system performance inspection device provided by the embodiments of the present application is described below. The system performance inspection device described below can be correspondingly referred to the system performance inspection method described above.

[0126] As shown in Figure 2 , the present application provides a system performance inspection device 200, which comprises:

[0127] The data acquisition module 201 is configured to acquire performance index data of a target system.

[0128] The trend analysis module 202 is configured to input the performance index data into a preset first model to obtain a preliminary analysis trend.

[0129] The knowledge retrieval module 203 is configured to determine a preset knowledge base, and perform retrieval in the knowledge base by using the preliminary analysis trend to obtain a best case and an analysis framework matched with the preliminary analysis trend.

[0130] The report generation module 204 is configured to input the performance indicator data, the best case and the analysis framework into a preset second model to obtain an inspection report of the target system.

[0131] In the above embodiment, the performance indicator data is first analyzed by the first model to replace the original fixed rule judgment, and the preliminary understanding of the performance indicator trend is realized. Then, the matching historical best case and analysis framework are searched in the knowledge base according to the preliminary analysis trend, and the experiential context and judgment standard are injected into the current decision. Finally, the performance indicator data, the best case and the analysis framework are input into the second model, and the decision-making process of the expert combining experience, method and real-time data is simulated by the second model. A highly accurate inspection report is obtained. In the above process, the deep analysis of the deep learning model and the knowledge base is combined to improve the understanding and judgment ability of the performance indicator fluctuation, thereby reducing the misjudgment probability and improving the accuracy and reliability of the performance inspection.

[0132] In one of the embodiments, the trend analysis module comprises:

[0133] The prompt word acquisition submodule is configured to acquire a preset prompt word.

[0134] The data input submodule is configured to input the prompt word and the performance indicator data into a preset first model.

[0135] The preliminary analysis submodule is configured to perform trend analysis on the performance indicator data according to the prompt word by the first model to identify the change mode and abnormal fluctuation of the performance indicator, and generate a preliminary analysis trend according to the change mode and abnormal fluctuation.

[0136] In one of the embodiments, the knowledge retrieval module comprises:

[0137] The report acquisition submodule is configured to acquire a historical inspection report.

[0138] The abstract extraction submodule is configured to perform structured processing on the historical inspection report to extract a performance data abstract.

[0139] The update submodule is configured to determine a preset knowledge base, and query a classification corresponding to the indicator type of the performance data abstract in the knowledge base, and then write the historical inspection report into the classification to update the knowledge base.

[0140] In one of the embodiments, the knowledge retrieval module comprises:

[0141] The keyword determination submodule is configured to analyze the preliminary analysis trend to determine a retrieval keyword.

[0142] The keyword searching submodule is configured to perform similarity matching in the knowledge base based on the searching keyword, and determine the best case in the knowledge base as the case with the highest similarity to the searching keyword.

[0143] The framework generating submodule is configured to extract the analysis steps and judgment criteria of the best case, and generate an analysis framework according to the analysis steps and judgment criteria.

[0144] In one of the embodiments, the report generating module comprises:

[0145] The data input submodule is configured to input the performance indicator data into the preset second model, and provide the best case and the analysis framework as context reference information to the second model.

[0146] The deep analysis submodule is configured to perform deep analysis on the performance indicator data by the second model and using the context reference information, and generate an inspection report comprising performance data summary, detailed analysis content and inspection conclusion.

[0147] In one of the embodiments, the deep analysis submodule comprises:

[0148] The rule determining unit is configured to determine a key judgment rule from the context reference information.

[0149] The data analysis unit is configured to analyze the performance indicator data according to the key judgment rule, and obtain the performance data summary, the detailed analysis content and the inspection conclusion, so as to generate the inspection report of the target system.

[0150] The key judgment rule comprises:

[0151] If the performance indicator data keeps below the preset threshold error rate in the latest continuous five days or more monitoring periods, it is determined that the target system has no degradation trend.

[0152] In one of the embodiments, the system performance inspection device further comprises:

[0153] The report analysis module is configured to analyze the inspection report of the target system, and if the analysis result is that the system is degraded, send the inspection report to the corresponding terminal.

[0154] The division of each module in the system performance inspection device is only used for illustration, and in other embodiments, the system performance inspection device can be divided into different modules as needed to complete all or part of the functions of the system performance inspection device. Each module in the system performance inspection device can be realized by software, hardware and combinations thereof. Each module can be embedded in or independent of the processor in the computer device in hardware form, or stored in the memory in the computer device in software form, so as to be called and executed by the processor to perform the operations corresponding to each module.

[0155] In one embodiment, the present application also provides a storage medium having stored computer readable instructions, which when executed by one or more processors, cause the one or more processors to perform the steps of the system performance patrol method according to any of the above embodiments.

[0156] In one embodiment, the present application also provides a computer device having stored computer readable instructions, which when executed by one or more processors, cause the one or more processors to perform the steps of the system performance patrol method according to any of the above embodiments.

[0157] As shown in Figure 3 , Figure 3 Fig. 3 is a schematic diagram of an internal structure of a computer device according to an embodiment of the present application. The computer device 300 can be provided as a server. As shown in Figure 3 , the computer device 300 includes a processing assembly 302, which further includes one or more processors, and a memory resource represented by a memory 301 for storing instructions, such as application programs, executable by the processing assembly 302. The application programs stored in the memory 301 can include one or more than one module each corresponding to a set of instructions. In addition, the processing assembly 302 is configured to execute the instructions to perform the system performance patrol method according to any of the above embodiments.

[0158] The computer device 300 can further include a power supply assembly 303 configured to perform power management of the computer device 300, a wired or wireless network interface 304 configured to connect the computer device 300 to a network, and an input output (I / O) interface 305. The computer device 300 can operate based on an operating system stored in the memory 301, such as Windows Server TM, Mac OS X TM, Unix TM, Linux TM, Free BSD TM, or the like.

[0159] Those skilled in the art can understand that Figure 3 the structure shown in the above figures is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or less components than those shown in the figures, or combine certain components, or have a different arrangement of components.

[0160] Finally, it should be noted that the terms "first" and "second", and the like, herein do not denote any order, quantity, combination or important / primary / secondary status, but are used to simply distinguish one element from another, and do not imply any actual relationship or sequence among those elements. Also, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. An element proceeded by "comprises... a" does not, without more constraints, exclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element. The terms "a" and "an" and "the" and similar referents in the context of this disclosure are to be construed to be inclusive of both singular and plural, unless the context clearly dictates otherwise. It will be further understood that the terms "comprises", "comprising", "has", "having", "includes" and / or "including", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can also include other elements not expressly listed or inherent to such process, method, article, or apparatus. The terms "and / or" and "or" as used herein refer to and encompass any and all possible combinations of one or more of the associated items, including but not limited to any one of the associated items alone, any two of the associated items, all of the associated items, and any permutation and combination thereof.

[0161] The various embodiments described in this specification are presented by way of example, and are not intended to limit the scope of the disclosure. Each embodiment is presented in a way that emphasizes its particular features, and the embodiments can be combined according to the needs of the user.

[0162] The above description of disclosed embodiments provides examples, and is not intended to limit the scope, applicability, or configuration of the disclosure. Various modifications can occur to those skilled in the art to which the disclosure relates, and such modifications are intended to be included within the scope of the disclosure. It is intended that the disclosure be construed as including all such modifications and equivalents. Accordingly, the application is not to be restricted, except in accordance with the following claims and their equivalents.

Claims

1. A method of system performance patrol, the method comprising: The method comprises: acquiring performance index data of a target system; inputting the performance index data into a preset first model to obtain a preliminary analysis trend; determining a preset knowledge base and searching the knowledge base using the preliminary analysis trend to obtain an optimal case and an analysis framework matching the preliminary analysis trend; inputting the performance index data, the optimal case and the analysis framework into a preset second model to obtain a patrol report of the target system.

2. The method of claim 1, wherein, The inputting of the performance index data into the preset first model to obtain the preliminary analysis trend comprises: acquiring a preset prompt word; inputting the prompt word and the performance index data into a preset first model; performing trend analysis on the performance index data according to the prompt word by the first model to identify a change mode and an abnormal fluctuation of the performance index, and generating a preliminary analysis trend according to the change mode and the abnormal fluctuation.

3. The method of claim 1, wherein, The updating process of the knowledge base comprises: acquiring historical patrol reports; performing structured processing on the historical patrol reports to extract performance data summaries; determining a preset knowledge base, querying a classification corresponding to an index type of the performance data summary in the knowledge base, and writing the historical patrol reports into the classification to update the knowledge base.

4. The method of claim 1, wherein, The searching of the knowledge base using the preliminary analysis trend to obtain an optimal case and an analysis framework matching the preliminary analysis trend comprises: analyzing the preliminary analysis trend to determine a search keyword; performing similarity matching in the knowledge base based on the search keyword, and determining a patrol case in the knowledge base having the highest similarity with the search keyword as the optimal case; extracting analysis steps and judgment criteria of the optimal case, and generating an analysis framework according to the analysis steps and the judgment criteria.

5. The method of claim 1, wherein, The inputting of the performance index data, the optimal case and the analysis framework into the preset second model to obtain the patrol report of the target system comprises: inputting the performance index data into a preset second model, and providing the optimal case and the analysis framework as context reference information to the second model; performing deep analysis on the performance index data by the second model and using the context reference information to generate a patrol report containing a performance data summary, detailed analysis content and a patrol conclusion.

6. The method of claim 5, wherein, The deep analysis of the performance index data using the context reference information to generate a patrol report containing a performance data summary, detailed analysis content and a patrol conclusion comprises: determining a key judgment rule from the context reference information; performing analysis on the performance index data according to the key judgment rule to obtain a performance data summary, detailed analysis content and a patrol conclusion to generate a patrol report of the target system; wherein the key judgment rule comprises: if the performance index data maintains an error rate lower than a preset threshold value in a monitoring period of five consecutive days or more, it is determined that the target system has no deterioration trend.

7. The method of claim 1 to 6, wherein, The method further comprises: Analyzing a patrol report of the target system, and if a patrol conclusion obtained by the analysis is system deterioration, sending the patrol report to a corresponding terminal.

8. A system performance patrol device characterized by comprising: The device comprises: a data acquisition module configured to acquire performance index data of a target system; a trend analysis module configured to input the performance index data into a preset first model to obtain a preliminary analysis trend; a knowledge retrieval module configured to determine a preset knowledge base and perform retrieval in the knowledge base by using the preliminary analysis trend to obtain a best case and an analysis framework matched with the preliminary analysis trend; a report generation module configured to input the performance index data, the best case and the analysis framework into a preset second model to obtain a patrol report of the target system.

9. A storage medium characterized by: The storage medium has computer readable instructions stored therein, and the computer readable instructions, when executed by one or more processors, cause the one or more processors to perform the steps of the system performance patrol method according to any one of claims 1 to 7.

10. A computer device, comprising: comprise: one or more processors, and a memory; the memory has computer readable instructions stored therein, and the computer readable instructions, when executed by the one or more processors, perform the steps of the system performance patrol method according to any one of claims 1 to 7.

Images