System performance monitoring method, device, equipment, medium and program product

Abstract

The invention provides a system performance monitoring method and device, equipment, a medium and a program product. The method comprises the steps of obtaining a to-be-processed target performance index value generated by a target system; determining a service scene mode corresponding to the target performance index value; according to the target performance index and the service scene mode, a threshold value type matched with the service scene mode is determined in a preset threshold value management rule base, the threshold value management rule base comprises threshold value types corresponding to different performance indexes in different service scene modes, and the different performance indexes comprise the target performance index; obtaining a performance index threshold value corresponding to the target performance index according to the threshold value type; and determining the performance of the target system according to a comparison result of the target performance index value and the performance index threshold value. According to the method, the performance index threshold most suitable for the current service state of the target system can be obtained, so that the current performance of the target system is accurately evaluated, and the defect existing when only a single performance index threshold is used is overcome.

Description

Technical Field

[0001] The present application relates to the field of information processing technologies, and in particular, to a system performance monitoring method, device, equipment, medium, and program product. Background Art

[0002] System performance monitoring is a key link to ensure the stable operation of the system. Currently, in related technologies, a single threshold monitoring mechanism is mainly used to monitor system performance. For example, only a fixed threshold monitoring mechanism or only a dynamic threshold monitoring mechanism is used to implement system performance monitoring.

[0003] However, the fixed threshold monitoring mechanism cannot adapt to the dynamic changes in system load and cannot ensure the accuracy of monitoring results. For example, during the business peak period, the system performance level significantly improves, resulting in the fixed threshold being easily triggered for false alarms, increasing the burden on operation and maintenance personnel. Secondly, the setting of the fixed threshold highly depends on the experience of operation and maintenance personnel and their understanding of the system. If set improperly, a large number of false alarms will also be generated. The dynamic threshold monitoring mechanism has a complex implementation process and high costs. The dynamic threshold algorithm requires continuous monitoring and adjustment, and at the same time consumes a large amount of computing resources and professional knowledge. Especially in large-scale or high-concurrency systems, the resource consumption of the dynamic threshold solution is particularly significant. Summary of the Invention

[0004] The present application provides a system performance monitoring method, device, equipment, medium, and program product to solve the defect of high error rate when only using a fixed threshold to implement the performance evaluation of the target system and the defect of high cost when only using a dynamic threshold to implement the performance evaluation of the target system in the prior art.

[0005] The present application provides a system performance monitoring method, including the following steps: Obtain the target performance metric value to be processed generated by the target system; Determine the business scenario mode corresponding to the target performance metric value, where the business scenario mode represents the business state presented by the target system during operation; According to the target performance metric and the business scenario mode, in the preset threshold management rule library, determine the threshold type that matches the business scenario mode. The threshold management rule library includes the threshold types corresponding to different performance metrics in different business scenario modes, and the different performance metrics include the target performance metric; According to the threshold type, obtain the performance metric threshold corresponding to the target performance metric; According to the comparison result between the target performance metric value and the performance metric threshold, determine the performance of the target system.

[0006] A system performance monitoring method provided by the present application, wherein the threshold types include fixed thresholds and dynamic thresholds; obtaining the performance metric threshold corresponding to the target performance metric according to the threshold type includes: If the threshold type is a fixed threshold, determine the performance metric threshold corresponding to the target performance metric from a plurality of pre-stored fixed thresholds; If the threshold type is a dynamic threshold, analyze the historical metric values corresponding to the target performance metric through time series data smoothing technology to obtain the performance metric threshold corresponding to the target performance metric.

[0007] A system performance monitoring method provided by the present application, wherein analyzing the historical metric values corresponding to the target performance metric through time series data smoothing technology to obtain the performance metric threshold corresponding to the target performance metric includes: According to the timestamps carried by the historical metric values, in all the historical metric values corresponding to the target performance metric, obtain a preset number of historical metric values whose timestamps are within a preset time range; Determine the average value of the preset number of historical metric values; Determine the performance metric threshold corresponding to the target performance metric according to the average value.

[0008] A system performance monitoring method provided by the present application, wherein according to the timestamps carried by the historical metric values, obtaining a preset number of historical metric values whose timestamps are within a preset time range in all the historical metric values corresponding to the target performance metric includes: According to the timestamps carried by the historical metric values, in all the historical metric values corresponding to the target performance metric after preprocessing operations, obtain the historical metric values whose timestamps are within the preset time range; Among the historical metric values whose timestamps are within the preset time range, obtain the latest preset number of historical metric values.

[0009] A system performance monitoring method provided by the present application, wherein determining the business scenario mode corresponding to the target performance metric value includes: Input the target performance metric, the target performance metric value, and the timestamp corresponding to the target performance metric value into a preset random forest model, and determine the business scenario mode corresponding to the target performance metric value according to the output result of the random forest model; Wherein, the random forest model is pre-trained based on the historical metric values corresponding to the target performance metric and the timestamps corresponding to each of the historical metric values to obtain an initial random forest model, and the random forest model is used to identify the business scenario modes corresponding to different performance metric values.

[0010] A system performance monitoring method provided by the present application, the fixed threshold corresponding to the target performance index is determined through the following steps: Obtain the historical index values corresponding to the target performance index, and sort the historical index values in ascending order; Divide the sorted historical index values into multiple different intervals, the multiple different intervals include a first interval, a second interval, and a third interval, and determine a first historical index value in the first interval, a second historical index value in the second interval, and a third historical index value in the third interval; Determine the fixed threshold corresponding to the target performance index according to the first historical index value, the second historical index value, and the third historical index value.

[0011] A system performance monitoring method provided by the present application, after determining the performance of the target system, the method further includes: If it is determined that the performance of the target system does not meet the preset requirements, generate an alarm message according to the target performance index value and the performance index threshold corresponding to the target performance index value; Output the alarm message.

[0012] The present application also provides a system performance monitoring device, including the following modules: A first acquisition module, configured to acquire the target performance index value to be processed generated by the target system; A first determination module, configured to determine the business scenario mode corresponding to the target performance index value, where the business scenario mode represents the business state presented by the target system during operation; A second determination module, configured to determine, according to the target performance index and the business scenario mode, a threshold type matching the business scenario mode in a preset threshold management rule library, where the threshold management rule library includes threshold types corresponding to different performance indexes in different business scenario modes, and the different performance indexes include the target performance index; A second acquisition module, configured to acquire the performance index threshold corresponding to the target performance index according to the threshold type; A third determination module, configured to determine the performance of the target system according to the comparison result between the target performance index value and the performance index threshold.

[0013] The present application also provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, it implements a system performance monitoring method as described in any one of the above.

[0014] The present application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements a system performance monitoring method as described in any one of the above.

[0015] The present application also provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements a system performance monitoring method as described in any one of the above.

[0016] The present application provides a system performance monitoring method, device, equipment, medium and program product. When implementing the system performance monitoring method of the present application, first obtain the target performance index value to be processed generated by the target system; then, determine the business scenario mode corresponding to the target performance index value, and then, according to the target performance index and the business scenario mode, in the preset threshold management rule library, determine the threshold type matching the business scenario mode, and according to the threshold type, obtain the performance index threshold corresponding to the target performance index; finally, according to the comparison result between the target performance index value and the performance index threshold, determine the performance of the target system. In the present application, by setting the threshold management rule to determine the threshold type corresponding to each target performance index, the performance index threshold most suitable for the current business state of the target system can be obtained, so as to accurately evaluate the current performance of the target system. On the one hand, this solution can avoid the phenomenon of high error rate when only using fixed thresholds to implement the performance evaluation of the target system, and on the other hand, it can also avoid the phenomenon of high cost when only using dynamic thresholds to implement the performance evaluation of the target system. Description of the Drawings

[0017] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a flowchart of a system performance monitoring method shown in an embodiment of the present application; Figure 2 It is a structural block diagram of a system performance monitoring device shown in an embodiment of the present application; Figure 3 It is a schematic physical structure diagram of an electronic device shown in an embodiment of the present application. Detailed Embodiments

[0019] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in this application will be clearly and completely described below with reference to the accompanying drawings in this application. Obviously, the described embodiments are some, but not all, of the embodiments of this application. All other embodiments obtained by those of ordinary skill in the art based on the embodiments in this application without creative efforts shall fall within the scope of protection of this application.

[0020] The execution subject of the system performance monitoring method of this application is a performance monitoring system, or a system performance monitoring device, or any type of electronic device. Below, taking the execution subject as a performance monitoring system as an example, a system performance monitoring method of this application will be described in detail.

[0021] Figure 1 It is a flowchart of a system performance monitoring method shown in an embodiment of this application. Refer to Figure 1 , the system performance monitoring method of this application specifically includes the following steps: Step 101, obtain the target performance index value to be processed generated by the target system.

[0022] Among them, the target system is the object to be monitored by the performance monitoring system. In this application, the performance monitoring system is responsible for real-time monitoring of various performance index values generated by the target system, so as to realize the performance evaluation of the target system.

[0023] The target system can be any type of system. For example, it can be a rail transit monitoring system in the rail transit industry, or various financial analysis systems in the financial industry, or various medical analysis systems in the medical industry. It can be specifically set according to actual needs, and the type of the target system is not specifically limited in this embodiment.

[0024] The target system usually includes at least one server. The performance monitoring system is used to monitor various performance index data generated during the operation of each server. A piece of performance index data includes a performance index, the performance index value corresponding to the performance index, and a timestamp, etc.

[0025] In this embodiment, there are various performance metrics for evaluating the performance of the target system, including CPU-related metrics (such as CPU usage rate, CPU load, CPU idle time), memory-related metrics (such as memory usage rate, available memory), disk I / O-related metrics (such as disk read / write rate, disk usage rate, disk I / O waiting time), system load and process metrics (such as system load average, number of active processes), application health metrics (such as application response time, application memory usage, application thread status, application error rate), network status metrics (such as network throughput, number of network connections, network error rate, network latency), and custom metrics. This embodiment does not specifically limit the types of performance metrics.

[0026] The target performance metric value refers to the specific value of the target performance metric. For example, when the target performance metric is CPU usage rate, the target performance metric value can be 30%; for another example, when the target performance metric is the number of active processes, the target performance metric value can be 10.

[0027] The target performance metric is any one of the various performance metrics used to evaluate the performance of the target system.

[0028] For the convenience of describing the method of this application, unless otherwise specified, each subsequent embodiment is described with the number of target performance metric values being one.

[0029] During the normal operation of the target system, for the same target performance metric, metric values will be generated at multiple different times. The performance monitoring system will monitor in real time whether each metric value meets the preset requirements. The target performance metric value to be processed refers to the performance metric value that has been generated by the target system and has not been processed by the performance monitoring system.

[0030] Step 102: Determine the business scenario mode corresponding to the target performance metric value, where the business scenario mode represents the business state presented by the target system during operation.

[0031] In this embodiment, the target system will present various different business states during operation, such as the normal operation state (the state in which the target system operates according to the established work process and parameters without any abnormalities or special events), being in the peak period (the state in which the operation load of the target system significantly increases due to increased business requirements during a specific time period), being in the maintenance window (the time period during which the target system is planned to be shut down for preventive maintenance, update and upgrade, or troubleshooting), the abnormal state (the unexpected and abnormal state that occurs during the operation of the target system, which may lead to performance degradation, failures, or security risks), etc.

[0032] The same target performance metric may produce different metric values in different business scenario modes. For example, for CPU utilization rate, it can correspond to one metric value (assumed to be 30%) in the normal operation state, and another metric value (assumed to be 60%) when the target system is at the peak period.

[0033] At a certain moment, a target performance metric value is generated from a business scenario mode in which the target system is located. For example, at the moment of 12:12:12 on XX / XX / XX, the CPU utilization rate (30%) cannot be generated from the four business scenario modes of normal operation state, peak period, maintenance window, and abnormal state at the same time, but only from one of these four business scenario modes. Therefore, in this embodiment, the business scenario mode corresponding to the target performance metric value refers to the business scenario mode in which the target system is located when generating the target performance metric value.

[0034] In this embodiment, the business scenario mode corresponding to the target performance metric value can be determined by any means. For example, the operation data of the target system when generating the target performance metric value can be intelligently analyzed to determine the business scenario mode in which the target system is located. It can also be determined by other means, which will be described later.

[0035] Step 103: According to the target performance metric and the business scenario mode, in the preset threshold management rule library, determine the threshold type that matches the business scenario mode. The threshold management rule library includes the threshold types corresponding to different performance metrics in different business scenario modes, and the different performance metrics include the target performance metric.

[0036] In this embodiment, the threshold types mainly include fixed thresholds and dynamic thresholds.

[0037] In this embodiment, the threshold management rule library includes multiple different performance metrics, and the threshold types corresponding to each performance metric in multiple different business scenario modes. The data recording situation in the threshold management rule library can be referred to as shown in Table 1 below:

[0038] Table 1 In this embodiment, one performance metric corresponds to one threshold type in one business scenario mode.

[0039] When performing step 103, according to the target performance metrics and the corresponding business scenario mode, the threshold type that matches the business scenario mode can be directly determined in the threshold management rule library. For example, for the target performance metric value 1 (the specific value of target performance metric 1), if its corresponding business scenario mode is business scenario mode 1, then by querying the threshold management rule library, it can be determined that the threshold type corresponding to the target performance metric value 1 is a fixed threshold; for another example, for the target performance metric value N (the specific value of target performance metric N), if its corresponding business scenario mode is business scenario mode M, then by querying the threshold management rule library, it can be determined that the threshold type corresponding to the target performance metric value N is a dynamic threshold.

[0040] In this embodiment, the information in the threshold management rule library is set according to the actual needs of the user.

[0041] Among them, the fixed threshold refers to the performance metric threshold determined before obtaining the target performance metric value to be processed, and the dynamic threshold refers to the performance metric threshold temporarily determined after obtaining the target performance metric value to be processed.

[0042] Step 104: Obtain the performance metric threshold corresponding to the target performance metric according to the threshold type.

[0043] In this embodiment, after determining the threshold type, the specific performance metric threshold can be further obtained. For example, for the target performance metric value of 30% (CPU usage rate), if the corresponding threshold type is a fixed threshold, the corresponding performance metric threshold is further obtained from the pre-stored fixed threshold information, such as 35%. For another example, for the target performance metric value of 30% (CPU usage rate), if the corresponding threshold type is a dynamic threshold, the corresponding dynamic threshold is temporarily generated to determine the performance metric threshold.

[0044] Step 105: Determine the performance of the target system according to the comparison result between the target performance metric value and the performance metric threshold.

[0045] Specifically, step 105 may include: If the comparison result is within the preset range, it is determined that the performance of the target system meets the standard; If the comparison result is not within the preset range, it is determined that the performance of the target system does not meet the standard.

[0046] After step 104, the target performance metric value can be compared with the performance metric threshold, and then check whether the comparison result is within the preset range, so as to determine whether the performance of the target system meets the standard.

[0047] Exemplarily, if the target performance metric is CPU usage rate, the target performance metric value is 30%, and the performance metric threshold is 40%, then the comparison result is that the target performance metric value is less than the performance metric threshold. Assuming the specified preset range is 0 <= CPU usage rate <= 40%, it means that the comparison result is within the preset range and the performance of the target system meets the standard. If the target performance metric value is 50%, then the comparison result is that the target performance metric value is greater than the performance metric threshold, and the comparison result is not within the preset range, so the performance of the target system does not meet the standard.

[0048] In this embodiment, the performance metric threshold corresponding to the target performance metric can be one or more.

[0049] Exemplarily, if the target performance metric is the number of active processes, the target performance metric value is 19, and the performance metric thresholds are 1 and 40, then the comparison result is that the target performance metric value is greater than 1 and less than 40. Assuming the specified preset range is 1 <= number of active processes <= 40, it means that the comparison result is within the preset range and the performance of the target system meets the standard. If the target performance metric value is 50, then the comparison result is that the target performance metric value is greater than the performance metric threshold 40, and the comparison result is not within the preset range, so the performance of the target system does not meet the standard.

[0050] To implement the system performance monitoring method of the present application, first obtain the target performance metric value to be processed generated by the target system; then, determine the business scenario mode corresponding to the target performance metric value, and then, according to the target performance metric and the business scenario mode, in the preset threshold management rule library, determine the threshold type that matches the business scenario mode, and according to the threshold type, obtain the performance metric threshold corresponding to the target performance metric; finally, determine the performance of the target system according to the comparison result between the target performance metric value and the performance metric threshold. In the present application, by setting the threshold management rule to determine the threshold type corresponding to each target performance metric, the performance metric threshold most suitable for the current business state of the target system can be obtained, so as to accurately evaluate the current performance of the target system. On the one hand, this solution can avoid the phenomenon of high error rate when only using fixed thresholds to evaluate the performance of the target system, and on the other hand, it can also avoid the phenomenon of high cost when only using dynamic thresholds to evaluate the performance of the target system.

[0051] Combined with the above embodiments, in one implementation, in step 104, according to the threshold type, obtaining the performance metric threshold corresponding to the target performance metric may specifically include: Step 1041: If the threshold type is a fixed threshold, determine the performance metric threshold corresponding to the target performance metric from multiple pre-stored fixed thresholds; Step 1042: If the threshold type is a dynamic threshold, analyze the historical indicator values corresponding to the target performance indicator through time series data smoothing technology to obtain the performance indicator threshold corresponding to the target performance indicator.

[0052] In this embodiment, the performance monitoring system will pre-determine and store the fixed thresholds corresponding to each performance indicator. Therefore, after determining that the threshold type corresponding to the target performance indicator value is a fixed threshold, the fixed threshold corresponding to the target performance indicator can be determined from the multiple pre-stored fixed thresholds as the performance indicator threshold corresponding to the target performance indicator value.

[0053] If it is determined that the threshold type corresponding to the target performance indicator value is a dynamic threshold, then it is necessary to temporarily generate the dynamic threshold corresponding to the target performance indicator as the performance indicator threshold corresponding to the target performance indicator value. Specifically, the process of generating the dynamic threshold includes: Step 1: Read the historical indicator values corresponding to the target performance indicator from the time series database.

[0054] Each historical indicator value carries a timestamp. The read historical indicator values can be all historical indicator values or the historical indicator values within a set time interval. The set time interval can be, for example, the most recent 3 months or the most recent 6 months, and the set time interval can be set according to actual needs.

[0055] Exemplarily, if the target performance indicator is the memory usage rate, then the specific values of the memory usage rate generated in the most recent 3 months can be read from the time series database.

[0056] Step 2: Analyze the historical indicator values obtained in Step 1 through time series data smoothing technology to obtain the performance indicator threshold corresponding to the target performance indicator.

[0057] Among them, time series data smoothing technology is a method used to eliminate or weaken the random fluctuations and noises in the data, thereby revealing the long-term trends and seasonal variations in the data, and can analyze the internal laws and potential patterns of the data. The simple moving average algorithm (SMA), weighted moving average method (WMA), and exponential moving average method (EMA) are commonly used methods in time series data smoothing technology. In Step 2, a suitable method can be adopted according to actual needs to determine the dynamic threshold. For example, the simple moving average algorithm can be used to analyze all historical indicator values, and the obtained dynamic threshold can be used as the performance indicator threshold corresponding to the target performance indicator.

[0058] In this embodiment, based on the historical index values corresponding to the target performance index, a dynamic threshold is obtained by using time series data smoothing technology, which can ensure that the obtained dynamic threshold is the performance index threshold most suitable for the current business state of the target system, and helps to improve the accuracy of the performance evaluation result of the target system.

[0059] Combined with the above embodiments, in one implementation, by analyzing the historical index values corresponding to the target performance index through time series data smoothing technology, the performance index threshold corresponding to the target performance index can include: According to the timestamps carried by the historical index values, among all the historical index values corresponding to the target performance index, obtain a preset number of historical index values whose timestamps are within a preset time range; Determine the average value of the preset number of historical index values; Determine the performance index threshold corresponding to the target performance index according to the average value.

[0060] In this embodiment, since the quantity stored in the time series database is large, if all the historical index values corresponding to the target performance index are analyzed, it will consume a large amount of computing resources and reduce the efficiency of performance evaluation of the target system. Therefore, among all the historical index values corresponding to the target performance index, the historical index values generated within a preset time range can be initially screened out, and then among the initially screened historical index values, a preset number of historical index values are further screened out.

[0061] Among them, how to screen out a preset number of historical index values from the initially screened historical index values can be determined according to a preset screening strategy. For example, the screening strategy can be to screen any preset number of historical index values, or to screen the latest preset number of historical index values, or to screen the preset number of historical index values located in the middle area of the preset time range.

[0062] In this embodiment, determining the performance index threshold corresponding to the target performance index according to the average value can include: Determine the average value as the performance index threshold corresponding to the target performance index; or Determine the product of the average value and the threshold coefficient as the performance index threshold corresponding to the target performance index.

[0063] The threshold coefficient can be set according to experience. Determining the product of the average value and the threshold coefficient as the performance index threshold corresponding to the target performance index can improve the accuracy of the determined performance index threshold, thereby improving the accuracy of the performance evaluation result.

[0064] Through this embodiment, the performance index threshold most suitable for the current business state of the target system can be obtained, which helps to improve the accuracy of the performance evaluation result of the target system.

[0065] In combination with the above embodiments, in one implementation, among all historical index values corresponding to the target performance index, obtaining a preset number of historical index values whose timestamps are within a preset time range may include: According to the timestamps carried by the historical index values, among all historical index values corresponding to the target performance index after the preprocessing operation, obtain the historical index values whose timestamps are within the preset time range; Among the historical index values whose timestamps are within the preset time range, obtain the latest preset number of historical index values.

[0066] In this embodiment, all data stored in the time series database are data after the preprocessing operation, that is, in the time series database, all historical index values corresponding to each performance index are historical index values after the preprocessing operation.

[0067] In specific implementation, first, among all historical index values corresponding to the target performance index after the preprocessing operation, preliminarily screen out the historical index values generated within the preset time range, and then among the preliminarily screened historical index values, use the latest generated preset number of historical index values as the preset number of historical index values finally used.

[0068] Among them, the preprocessing operation includes but is not limited to: data cleaning (processing missing values, outliers, duplicate data, etc.), format conversion (converting data into a format suitable for analysis or storage, such as adjusting the format of timestamps, converting data into specific units, etc.). The purpose of the preprocessing operation is to improve data quality, so any preprocessing operation that can improve data quality can be used in this application.

[0069] In the related art, when using a dynamic threshold to implement performance monitoring, historical data is not preprocessed. Once there are deviations or anomalies in the historical data, the accuracy of the dynamic threshold will be reduced, and further the stability of the performance monitoring result will be reduced. However, in this embodiment, a preset number of historical index values after the preprocessing operation are used to determine the dynamic threshold, which can overcome the problems in the related art and ensure the accuracy of the finally determined dynamic threshold. Secondly, using the latest preset number of historical index values to determine the dynamic threshold can further improve the accuracy of the determined dynamic threshold, and thus ensure the accuracy of the performance evaluation effect.

[0070] In this embodiment, the time series database stores sufficiently rich historical data by default. When the performance monitoring system is just put into use, there may be a situation where historical data is insufficient. At this time, these problems can be avoided by modifying the threshold management rule library. For example, the threshold types corresponding to different performance metrics in different business scenario modes can be modified to fixed thresholds, and the initial values of these fixed thresholds can be determined according to manual experience. After sufficiently rich historical data is stored in the time series database, the information in the threshold management rule library is adjusted again according to the historical data, and the initial values of the fixed thresholds are adjusted again according to the historical data. After that, the initial values of each fixed threshold can be updated according to the historical data stored in the time series database at preset time intervals, so as to improve the accuracy of the fixed thresholds and the accuracy of the performance evaluation results. Combining the above embodiments, in one implementation manner, in step 102, determining the business scenario mode corresponding to the target performance metric value may specifically include: Input the target performance metric, the target performance metric value, and the time stamp corresponding to the target performance metric value into a preset random forest model, and determine the business scenario mode corresponding to the target performance metric value according to the output result of the random forest model; Among them, the random forest model is obtained by training the initial random forest model in advance according to the historical metric values corresponding to the target performance metric and the time stamps corresponding to each historical metric value. The random forest model is used to identify the business scenario modes corresponding to different performance metric values.

[0071] In this embodiment, the random forest model is an ensemble learning method that improves the accuracy and robustness of prediction by constructing multiple decision trees and summarizing their prediction results. The training process of the random forest model in this application that can identify the business scenario modes corresponding to different performance metric values includes: Step1: Data preparation. Obtain the preprocessed data used for training the model from the time series database. This part of the data includes the historical metric values corresponding to various different performance metrics.

[0072] Step2: Feature selection. Based on business requirements and data characteristics, select or create features that are helpful for identifying business scenario modes. Feature selection is not just a simple feature extraction of the data obtained in Step1, but an analysis and processing of the obtained data to obtain metrics that can reflect different business states of the target system. For example, the data obtained in Step1 includes multiple historical metric values corresponding to performance metrics such as CPU usage rate and memory usage rate, and the feature selection process may involve calculating the change rate, average value, peak value, etc. of these performance metrics, so as to obtain features that can better reflect the business state of the target system.

[0073] Step 3: Data tagging. Assign a tag representing the business scenario pattern to each piece of data, which can be obtained specifically through manual annotation or automatically based on the existing knowledge base.

[0074] Step 4: Divide the training set and the test set. Divide the data obtained in Step 3 into a training set and a test set. The training set is used to train the random forest model, and the test set is used to evaluate the performance of the trained random forest model.

[0075] Step 5: Train the random forest model. First, construct decision trees. For each tree in the random forest, randomly draw a sample subset from the training set. At each node of each decision tree, randomly select a part of the features (usually about the square root of the total number of features), and then select the best splitting point based on these features. Repeat this process until reaching the maximum depth of the tree or the number of samples in the node is less than a certain threshold. Then, summarize the prediction results of all decision trees. For classification problems in this application (such as the recognition of business scenario patterns), majority voting can be used to determine the final recognition result.

[0076] Step 6: Model tuning and validation.

[0077] In Step 6, the time series cross-validation method can be used to evaluate the model performance. Time series data has time dependence, and this method can ensure that the model performs well on unseen data, thus avoiding the problems of overfitting or underfitting.

[0078] In Step 6, during the process of using the time series cross - validation method for cross - validation, metrics such as Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Symmetric Mean Absolute Percentage Error (sMAPE) can be used to measure the accuracy of the model's prediction results on each validation set. When using the mean absolute error for evaluation, the average of the absolute values of the differences between the predicted values and the true values is calculated. This value can reflect the magnitude of the prediction error, and the smaller the value, the better the model performance. When using the root mean square error for evaluation, the square root of the average of the squares of the differences between the predicted values and the true values is calculated. Compared with the mean absolute error, the root mean square error penalizes larger errors more severely, so this metric value is more sensitive to outliers. The symmetric mean absolute percentage error is a percentage error metric that considers the relative difference between the predicted value and the true value. The smaller the value of this metric, the higher the prediction accuracy of the model. Therefore, if the model performs poorly during a certain time period (such as peak hours or maintenance windows), the model parameters or structure can be adjusted according to the validation results. When adjusting the parameters, the parameters of the random forest, such as the number of trees, maximum depth, minimum sample split number, etc., can be adjusted to optimize the model performance.

[0079] In this embodiment, after obtaining the trained random forest model, the target performance metric, the target performance metric value, and the timestamp corresponding to the target performance metric value are input into the trained random forest model, and the random forest model can output the recognition result, that is, the business scenario pattern corresponding to the target performance metric value.

[0080] In this embodiment, by using the random forest algorithm to determine the business scenario pattern corresponding to the target performance metric value, the current business state of the target system can be accurately identified, which helps to subsequently accurately determine the performance metric threshold corresponding to the target performance metric and improve the accuracy of the performance evaluation result of the target system.

[0081] Combined with the above embodiments, in one implementation, the fixed threshold corresponding to the target performance metric is determined in advance through the following steps: Obtain the historical metric values corresponding to the target performance metric and sort the historical metric values in ascending order; Divide the sorted historical indicator values into multiple different intervals, where the multiple different intervals include a first interval, a second interval, and a third interval. Determine a first historical indicator value in the first interval, a second historical indicator value in the second interval, and a third historical indicator value in the third interval. Then, determine a fixed threshold corresponding to the target performance indicator based on the first historical indicator value, the second historical indicator value, and the third historical indicator value.

[0082] In this embodiment, when obtaining the historical indicator values corresponding to the target performance indicator, the historical indicator values within a set time range (set according to actual needs) in the time series database can be obtained. For example, the historical indicator values generated in the most recent three months or the most recent six months can be obtained.

[0083] Among them, each obtained historical indicator value is also data after preprocessing operations to ensure the accuracy of the finally determined fixed threshold.

[0084] After obtaining the historical indicator values, sort all the historical indicator values in ascending order according to the historical indicator values from small to large. Then, divide the sorted historical indicator values into multiple intervals, including a first interval, a second interval, and a third interval. Next, determine a first historical indicator value in the first interval, a second historical indicator value in the second interval, and a third historical indicator value in the third interval. Among them, the first interval, the second interval, and the third interval can be determined from the multiple divided intervals according to actual needs. When obtaining the first historical indicator value, the second historical indicator value, and the third historical indicator value, values can also be taken within the corresponding intervals according to actual needs. This embodiment does not specifically limit the division method of each interval and the obtaining method of each historical indicator value within the corresponding interval.

[0085] In one implementation, all the historical metric values after ascending sorting can be evenly divided into 4 portions of data. Then, in the order of the historical metric values from small to large, the first portion of data is used as the data within the first interval, the second portion of data is used as the data within the second interval, and the third portion of data is used as the data within the third interval. Subsequently, the maximum value in the first portion of data can be used as the first historical metric value, the maximum value in the second portion of data can be used as the second historical metric value, and the maximum value in the third portion of data can be used as the third historical metric value. In this way, the first historical metric value is the maximum value among the historical metric values in the first 25% of the historical metric values after ascending sorting, the second historical metric value is the maximum value among the historical metric values in the first 50% of the historical metric values after ascending sorting, and the third historical metric value is the maximum value among the historical metric values in the first 75% of the historical metric values after ascending sorting. By way of example, if there are 100 historical metric values after ascending sorting in total, the finally determined first historical metric value is the 25th historical metric value, the second historical metric value is the 50th historical metric value, and the third historical metric value is the 75th historical metric value.

[0086] In another implementation, all the historical metric values after ascending sorting can be evenly divided into 4 portions of data. Then, in the order of the historical metric values from small to large, the second portion of data is used as the data within the first interval, the third portion of data is used as the data within the second interval, and the fourth portion of data is used as the data within the third interval. Subsequently, the minimum value in the second portion of data can be used as the first historical metric value, the minimum value in the third portion of data can be used as the second historical metric value, and the minimum value in the fourth portion of data can be used as the third historical metric value. In this way, the first historical metric value is the minimum value among the historical metric values greater than the historical metric values in the first 25% of the historical metric values after ascending sorting. The second historical metric value is the minimum value among the historical metric values greater than the historical metric values in the first 50% of the historical metric values after ascending sorting. The third historical metric value is the minimum value among the historical metric values greater than the historical metric values in the first 75% of the historical metric values after ascending sorting. By way of example, if there are 100 historical metric values after ascending sorting in total, the finally determined first historical metric value is the 26th historical metric value, the second historical metric value is the 51st historical metric value, and the third historical metric value is the 76th historical metric value.

[0087] In the above two ways, dividing all the historical metric values into multiple intervals and ignoring the historical metric values in the head and tail intervals when obtaining the first to third historical metric values can avoid the influence of extreme historical metric values on the fixed threshold and improve the accuracy of the fixed threshold.

[0088] Next, according to the first historical index value, the second historical index value, and the third historical index value, a fixed threshold corresponding to the target performance index can be determined. For example, when using Q1, Q2, and Q3 to represent the first historical index value, the second historical index value, and the third historical index value in sequence, the fixed threshold can be [Q3 + 1.5 * (Q3 - Q1)] or [Q1 - 1.5 * (Q3 - Q1)]. In actual application, [Q3 + 1.5 * (Q3 - Q1)] can be used as the upper limit value and [Q1 - 1.5 * (Q3 - Q1)] can be used as the lower limit value, that is, values less than [Q1 - 1.5 * (Q3 - Q1)] or greater than [Q3 + 1.5 * (Q3 - Q1)] are all outlier values. Of course, how to set the outlier range according to the fixed threshold can be determined according to the actual business requirements.

[0089] Through this embodiment, the accuracy of the fixed threshold corresponding to different target performance indexes can be improved, and further the accuracy of the performance evaluation result of the target system can be improved.

[0090] Combined with the above embodiments, in one implementation manner, after determining the performance of the target system, the method of the present application may further include: If it is determined that the performance of the target system does not meet the preset requirements, generate an alarm message according to the target performance index value and the performance index threshold corresponding to the target performance index value; Output the alarm message.

[0091] In this embodiment, if the performance of the target system does not meet the preset requirements, the performance monitoring system automatically generates and outputs an alarm message to timely remind relevant operation and maintenance personnel to check the target system.

[0092] In one implementation manner, in addition to the target performance index value and the performance index threshold corresponding to the target performance index value, the alarm message may further include the business scenario mode corresponding to the target performance index value, etc., so as to facilitate relevant operation and maintenance personnel to timely check whether the setting of the performance index threshold is reasonable, and readjust the performance index threshold when it is unreasonable, so as to ensure the accuracy of the subsequent performance evaluation result of the target system.

[0093] Next, a complete embodiment will be used to illustrate the system performance monitoring method of the present application. This embodiment includes a pre-stage and a real-time monitoring stage, and specifically includes the following steps: (1) Pre-stage: Step 1: Deploy components such as Prometheus, IoTDB, Grafana, Ansible, and AlertManager in the performance monitoring system to ensure that each component runs normally.

[0094] Step 2: Prometheus is responsible for monitoring multiple servers in the target system and collecting various performance metric data (including performance metrics, performance metric values, and timestamps) generated by each server.

[0095] Step 3: Set the data source in Grafana, connect to Prometheus, create a Dashboard to query a certain performance metric data using PromQL statements, and analyze and display the queried performance metric data.

[0096] Step 4: Perform preprocessing operations on all the collected performance metric data. Specifically, clean all the collected performance metric data and convert all the collected performance metric data into a suitable format to improve data quality. Finally, store the preprocessed performance metric data in the IoTDB time series database.

[0097] Among them, Prometheus is a system monitoring and alerting tool suite suitable for cloud-native environments. It collects time series data from configured targets (the target system in this embodiment) through the HTTP protocol. By configuring the scrape_configs section of Prometheus, multiple servers (or applications) in the target system can be specified as monitoring targets. Prometheus pulls various performance metric data from the configured targets based on the configured time interval, and the performance metric data are all time series data.

[0098] Grafana is a platform-independent analysis and interactive visualization software that supports functions such as querying, visualization, and alerting. In Grafana, after adding Prometheus as the data source, by configuring the data source URL of Prometheus and other necessary authentication information, Grafana can connect to Prometheus and access all the data collected by Prometheus.

[0099] Dashboard is a concept in Grafana that allows users to organize and display multiple panels, and each panel can display the results of one or more queries.

[0100] PromQL (Prometheus Query Language) is the query language of Prometheus used to extract data from the time series database of Prometheus. Using the chart and visualization tools of Grafana, users can intuitively view and analyze the data collected from Prometheus.

[0101] IoTDB (Internet of Things Database) is a time-series database designed specifically for Internet of Things (IoT) and big data applications. The IoTDB time-series database features efficient data compression, fast query performance, and the ability to support complex queries.

[0102] Ansible is an automation platform whose core functions are implemented through playbooks (files in YAML format). Playbooks define the tasks to be executed, the targets, the execution order, and the dependencies between tasks. After determining the fixed thresholds or dynamic thresholds, Ansible can apply these fixed thresholds to the performance monitoring system, such as possibly involving updating configuration files, restarting services, or sending commands to the performance monitoring system to apply new performance metric thresholds.

[0103] AlertManager is part of Prometheus and is responsible for handling alerts from Prometheus. When Prometheus detects that the comparison result between a certain performance metric value and the corresponding performance metric threshold does not meet the preset result, it generates an alert and sends the alert to AlertManager.

[0104] (2) Real-time monitoring stage Step 1: Obtain a target performance metric value currently generated by the target system. For example, the target performance metric value could be 30% (CPU usage rate); Step 2: Determine the business scenario pattern corresponding to the target performance metric value through a pre-trained random forest model; Step 3: In the preset threshold management rule library, determine the threshold type that matches the business scenario pattern. If the threshold type is a fixed threshold, determine the performance metric threshold corresponding to the target performance metric from multiple pre-stored fixed thresholds; if the threshold type is a dynamic threshold, analyze the historical metric values corresponding to the target performance metric through time series data smoothing technology to obtain the performance metric threshold corresponding to the target performance metric; Step 4: Determine the performance of the target system based on the comparison result between the target performance metric value and the performance metric threshold.

[0105] This application proposes a new threshold management mechanism that cleverly combines fixed thresholds and dynamic thresholds to achieve more accurate alarms. The new threshold management mechanism has at least the following technical effects: (1) Avoids the phenomenon of high error rate when only using fixed thresholds to evaluate the performance of the target system (which cannot fully adapt to the load changes of the target system, and the setting of fixed thresholds depends on the experience of operation and maintenance personnel).

[0106] (2) It avoids the phenomenon of high cost when only using dynamic thresholds to evaluate the performance of the target system.

[0107] (3) Both the dynamic threshold and the fixed threshold use pre-processed historical data, which can ensure the data quality, improve the accuracy of the performance metric threshold, and avoid unstable performance monitoring results.

[0108] (4) In this application, by setting threshold management rules to determine the threshold type corresponding to each target performance metric, the performance metric threshold most suitable for the current business state of the target system can be obtained, so as to accurately evaluate the current performance of the target system, enhancing the reliability and stability of the target system. Next, a system performance monitoring device provided by this application will be described. The system performance monitoring device described below can be correspondingly referred to the system performance monitoring method described above.

[0109] Figure 2 It is a structural block diagram of a system performance monitoring device shown in an embodiment of this application. Referring to Figure 2 , the system performance monitoring device 200 of this application includes: A first acquisition module 201, configured to acquire target performance metric values to be processed generated by the target system; A first determination module 202, configured to determine the business scenario mode corresponding to the target performance metric value, where the business scenario mode represents the business state presented by the target system during operation; A second determination module 203, configured to determine, according to the target performance metric and the business scenario mode, a threshold type matching the business scenario mode in a preset threshold management rule library, where the threshold management rule library includes threshold types corresponding to different performance metrics in different business scenario modes, and the different performance metrics include the target performance metric; A second acquisition module 204, configured to acquire the performance metric threshold corresponding to the target performance metric according to the threshold type; A third determination module 205, configured to determine the performance of the target system according to the comparison result between the target performance metric value and the performance metric threshold.

[0110] According to a system performance monitoring device 200 provided by this application, the threshold type includes a fixed threshold and a dynamic threshold; the second acquisition module 204 includes: A first acquisition sub-module, configured to, if the threshold type is a fixed threshold, determine the performance metric threshold corresponding to the target performance metric from multiple pre-stored fixed thresholds; The second acquisition sub-module is configured to, if the threshold type is a dynamic threshold, analyze the historical index values corresponding to the target performance index through a time series data smoothing technique to obtain the performance index threshold corresponding to the target performance index.

[0111] According to a system performance monitoring device 200 provided by the present application, the second acquisition sub-module includes: The third acquisition sub-module is configured to, according to the timestamps carried by the historical index values, obtain a preset number of historical index values whose timestamps are within a preset time range among all the historical index values corresponding to the target performance index; The first determination sub-module is configured to determine the average value of the preset number of historical index values; The second determination sub-module is configured to determine the performance index threshold corresponding to the target performance index according to the average value.

[0112] According to a system performance monitoring device 200 provided by the present application, the third acquisition sub-module includes: The fourth acquisition sub-module is configured to, according to the timestamps carried by the historical index values, obtain the historical index values whose timestamps are within the preset time range among all the historical index values corresponding to the target performance index after a preprocessing operation; The fifth acquisition sub-module is configured to obtain the latest preset number of historical index values among the historical index values whose timestamps are within the preset time range.

[0113] According to a system performance monitoring device 200 provided by the present application, the first determination module 201 includes: The third determination sub-module is configured to input the target performance index, the target performance index value, and the timestamp corresponding to the target performance index value into a preset random forest model, and determine the business scenario mode corresponding to the target performance index value according to the output result of the random forest model; Wherein, the random forest model is pre-trained from an initial random forest model according to the historical index values corresponding to the target performance index and the timestamps corresponding to each of the historical index values, and the random forest model is used to identify the business scenario modes corresponding to different performance index values.

[0114] According to a system performance monitoring device 200 provided by the present application, it further includes: The third acquisition module is configured to obtain the historical index values corresponding to the target performance index and sort the historical index values in ascending order; A fourth determination module, configured to divide the sorted historical metric values into multiple different intervals, where the multiple different intervals include a first interval, a second interval, and a third interval, and determine a first historical metric value in the first interval, a second historical metric value in the second interval, and a third historical metric value in the third interval; A fifth determination module, configured to determine a fixed threshold corresponding to the target performance metric according to the first historical metric value, the second historical metric value, and the third historical metric value.

[0115] A system performance monitoring device 200 provided by the present application further includes: A generation module, configured to, after determining the performance of the target system, if it is determined that the performance of the target system does not meet the preset requirements, generate an alarm message according to the target performance metric value and the performance metric threshold corresponding to the target performance metric value; An output module, configured to output the alarm message.

[0116] Figure 3 is a schematic physical structure diagram of an electronic device shown in an embodiment of the present application. As Figure 3 shown, the electronic device may include: a processor 310, a communication interface 320, a memory 330, and a communication bus 340. Among them, the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call logical instructions in the memory 330 to execute a system performance monitoring method, and the method includes: Obtain a target performance metric value to be processed generated by the target system; Determine a service scenario mode corresponding to the target performance metric value, where the service scenario mode represents a service state presented by the target system during operation; According to the target performance metric and the service scenario mode, in a preset threshold management rule library, determine a threshold type that matches the service scenario mode. The threshold management rule library includes threshold types corresponding to different performance metrics in different service scenario modes, and the different performance metrics include the target performance metric; According to the threshold type, obtain a performance metric threshold corresponding to the target performance metric; Determine the performance of the target system according to a comparison result between the target performance metric value and the performance metric threshold.

[0117] In addition, when the logical instructions in the above-mentioned memory 330 can be implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several 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 methods described in various embodiments of this application. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical discs that can store program codes.

[0118] On the other hand, this application also provides a computer program product. The computer program product includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute a system performance monitoring method provided by the above-mentioned various methods. The method includes: obtaining a target performance index value to be processed generated by a target system; determining a service scenario mode corresponding to the target performance index value, where the service scenario mode represents the service state presented by the target system during operation; According to the target performance index and the service scenario mode, in a preset threshold management rule library, determining a threshold type that matches the service scenario mode. The threshold management rule library includes threshold types corresponding to different performance indexes in different service scenario modes, and the different performance indexes include the target performance index; According to the threshold type, obtaining a performance index threshold corresponding to the target performance index; Determining the performance of the target system according to the comparison result between the target performance index value and the performance index threshold.

[0119] On another aspect, this application also provides a non-transitory computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, it implements a system performance monitoring method provided by the above-mentioned various methods. The method includes: obtaining a target performance index value to be processed generated by a target system; determining a service scenario mode corresponding to the target performance index value, where the service scenario mode represents the service state presented by the target system during operation; According to the target performance indicator and the business scenario mode, in a preset threshold management rule library, determine the threshold type that matches the business scenario mode. The threshold management rule library includes the threshold types corresponding to different performance indicators in different business scenario modes, and the different performance indicators include the target performance indicator; According to the threshold type, obtain the performance indicator threshold corresponding to the target performance indicator; According to the comparison result between the target performance indicator value and the performance indicator threshold, determine the performance of the target system.

[0120] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed to multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art can understand and implement it without creative work.

[0121] Through the description of the above embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution, in essence, or the part that contributes to the prior art can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disc, etc., and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0122] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, and are not intended to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of each embodiment of the present application.

Claims

1. A system performance monitoring method, characterized in that: include: Obtaining the target performance indicator value to be processed generated by the target system; Determine a business scenario mode corresponding to the target performance indicator value, wherein the business scenario mode represents a business state presented by the target system during operation; According to the target performance indicator and the business scenario mode, a threshold type matching the business scenario mode is determined in a preset threshold management rule base, wherein the threshold management rule base includes threshold types corresponding to different performance indicators in different business scenario modes, and the different performance indicators include the target performance indicator; According to the threshold type, obtaining a performance indicator threshold corresponding to the target performance indicator; The performance of the target system is determined according to a comparison result of the target performance indicator value and the performance indicator threshold.

2. The system performance monitoring method according to claim 1, characterized in that: The threshold type includes a fixed threshold and a dynamic threshold; and obtaining the performance indicator threshold corresponding to the target performance indicator according to the threshold type includes: If the threshold type is a fixed threshold, determining a performance indicator threshold corresponding to the target performance indicator from a plurality of pre-stored fixed thresholds; If the threshold type is a dynamic threshold, the historical indicator value corresponding to the target performance indicator is analyzed by time series data smoothing technology to obtain the performance indicator threshold corresponding to the target performance indicator.

3. The system performance monitoring method according to claim 2, characterized in that: The analyzing the historical indicator value corresponding to the target performance indicator by using the time series data smoothing technology to obtain the performance indicator threshold corresponding to the target performance indicator includes: According to the timestamps carried by the historical indicator values, a preset number of historical indicator values ​​whose timestamps are within a preset time range are obtained from all the historical indicator values ​​corresponding to the target performance indicator; Determine an average value of the preset number of historical indicator values; A performance indicator threshold corresponding to the target performance indicator is determined according to the average value.

4. The system performance monitoring method according to claim 3, characterized in that: The acquiring, according to the timestamps carried by the historical indicator values, a preset number of historical indicator values ​​whose timestamps are within a preset time range from among all the historical indicator values ​​corresponding to the target performance indicator, comprises: According to the timestamps carried by the historical indicator values, obtaining, from all the historical indicator values ​​corresponding to the target performance indicator after the preprocessing operation, historical indicator values ​​whose timestamps are within the preset time range; Among the historical indicator values ​​whose timestamps are within the preset time range, the latest preset number of historical indicator values ​​are obtained.

5. The system performance monitoring method according to claim 1, characterized in that: The determining of the business scenario mode corresponding to the target performance indicator value includes: Inputting the target performance indicator, the target performance indicator value, and the timestamp corresponding to the target performance indicator value into a preset random forest model, and determining the business scenario mode corresponding to the target performance indicator value according to the output result of the random forest model; Among them, the random forest model is obtained by pre-training an initial random forest model based on the historical indicator values ​​corresponding to the target performance indicator and the timestamps corresponding to each of the historical indicator values. The random forest model is used to identify business scenario patterns corresponding to different performance indicator values.

6. The system performance monitoring method according to claim 1, characterized in that: The fixed threshold corresponding to the target performance indicator is determined by the following steps: Obtaining historical indicator values ​​corresponding to the target performance indicator, and arranging the historical indicator values ​​in ascending order; Dividing the historical indicator values ​​after ascending order into a plurality of different intervals, the plurality of different intervals including a first interval, a second interval, and a third interval, and determining a first historical indicator value in the first interval, determining a second historical indicator value in the second interval, and determining a third historical indicator value in the third interval; A fixed threshold corresponding to the target performance indicator is determined according to the first historical indicator value, the second historical indicator value, and the third historical indicator value.

7. The system performance monitoring method according to claim 1, characterized in that: After determining the performance of the target system, the method further includes: If it is determined that the performance of the target system does not meet the preset requirements, generating alarm information according to the target performance indicator value and the performance indicator threshold corresponding to the target performance indicator value; Output the warning information.

8. A system performance monitoring device, characterized in that: include: A first acquisition module is used to acquire a target performance indicator value to be processed generated by a target system; A first determination module is used to determine a business scenario mode corresponding to the target performance indicator value, wherein the business scenario mode represents a business state presented by the target system during operation; A second determination module is used to determine, according to the target performance indicator and the business scenario mode, a threshold type matching the business scenario mode in a preset threshold management rule base, wherein the threshold management rule base includes threshold types corresponding to different performance indicators in different business scenario modes, and the different performance indicators include the target performance indicator; A second acquisition module, configured to acquire a performance indicator threshold corresponding to the target performance indicator according to the threshold type; The third determination module is used to determine the performance of the target system according to the comparison result of the target performance indicator value and the performance indicator threshold.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the system performance monitoring method according to any one of claims 1 to 7 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, a system performance monitoring method as described in any one of claims 1 to 7 is implemented.

11. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, a system performance monitoring method as described in any one of claims 1 to 7 is implemented.

Images