Method and system for monitoring application performance of linux operating system

Abstract

The invention discloses an application performance monitoring method and system of a linux operating system, and the method comprises the following steps: continuously monitoring a performance index of a target application, tracking a key performance index of a system in a full-link manner, and monitoring a full life cycle related to business data; if the performance index of the target application is abnormal, triggering an alarm, recording an abnormal total system index, and recording an alarm log; intelligent analysis and pre-judgment are carried out on full-amount performance index data, prediction analysis is carried out on performance data through a machine learning algorithm, and possible performance problems are found in advance; and intelligent automatic tuning is realized in combination with expert experience. According to the method, the application performance indexes and the system indexes are monitored in real time, the abnormal results of the application performance indexes and the system indexes are integrated to obtain final performance index abnormal information and record the performance index abnormal information, the problem checking and solving process of a system administrator is greatly simplified, and meanwhile automation and intelligentization of operation and maintenance work are further promoted through intelligent analysis and intelligent adjustment and optimization functions.

Description

Technical Field

[0001] The present invention relates to computer software technology, and in particular to an application performance monitoring method and system for a Linux operating system. Background Art

[0002] In a complex IT environment, applications may be affected by a variety of factors, such as network latency, hardware failure, resource contention, code defects, etc., which may lead to application performance degradation. If application performance anomalies are not discovered in time, it will be difficult to find the cause of the anomaly, or even cause application failure. The performance of the application is directly related to user experience, business efficiency, and the overall competitiveness of the enterprise.

[0003] When performing application performance monitoring, existing technical solutions need to collect statistics and analyze logs of a large number of system logs and logs of various components in the application software stack to monitor application performance. For example, the patent application with publication number CN116257416A proposes an application performance monitoring system and method, storage medium, and electronic device, which solves the problem of inaccurate detection of application performance and achieves the effect of accurately detecting application performance. The system includes: a file management module for obtaining logs of multiple applications, wherein the logs of multiple applications include at least one of the following: database logs, middleware logs, probe logs, event logs, status code logs, interface logs, and link logs; a processing module for determining the performance of multiple applications based on the logs of multiple applications and generating performance reports for multiple applications; a display device for displaying logs and performance reports of multiple applications. This solution has the following defects:

[0004] (1) Analyze application performance based only on application logs. If the log collection is incomplete or contains errors, the performance report generated based on these logs may be inaccurate and cannot truly reflect the performance status of the application.

[0005] (2) Monitoring system resources is not considered. Application logs usually do not include the current system status, making it extremely difficult to analyze the application.

[0006] (3) To monitor the overall performance of an application, users need to select a performance indicator of interest. Abnormal monitoring of the indicator values ​​of the application is not supported.

[0007] To sum up, the existing application performance monitoring methods need to analyze a large number of log files. At the same time, the application performance indicators need to be analyzed after these logs to obtain the monitoring results. There are problems such as low efficiency, long time consumption, and strong dependence of performance monitoring results on log files. In addition, if you want to track the cause of application performance anomalies, you need to manually review and analyze the above log files. In scenarios with large log volumes, it takes a lot of time to analyze. Summary of the invention

[0008] The technical problem to be solved by the present invention is as follows: In view of the above-mentioned problems in the prior art, a method and system for monitoring application performance of a Linux operating system are provided to monitor application performance indicators and system indicators in real time, and to obtain and record the abnormal information of the final performance indicators by combining the abnormal results of the two, thereby greatly simplifying the problem troubleshooting and solving process of the system administrator, providing strong technical support for the system administrator, and being able to deal with application performance problems more easily and accurately.

[0009] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0010] A method for monitoring application performance of a Linux operating system comprises the following steps:

[0011] Continuously monitor the performance indicators of the target application and track the key performance indicators of the system throughout the entire link, including network throughput, CPU utilization, memory usage, disk IO, etc., monitoring the entire life cycle of business data;

[0012] If the performance indicators of the target application are monitored to be abnormal, an alarm is triggered immediately, the corresponding abnormal key performance indicators of the entire system are recorded, and the abnormal performance indicators of the target application and the corresponding abnormal key performance indicators of the entire system are added to the alarm log;

[0013] Intelligent analysis and prediction: intelligent analysis and prediction of all performance indicator data to obtain the prediction results of performance indicator data;

[0014] Intelligent automatic tuning automatically adjusts resource allocation based on the predicted results of performance indicator data and combined with expert experience to achieve performance optimization.

[0015] Furthermore, when continuously monitoring the performance indicators of the target application, the following steps are included:

[0016] Parse the application monitoring configuration file to obtain the performance indicator name, performance indicator acquisition path and corresponding threshold of the target application;

[0017] If the parsing of the application configuration file fails, the process ends and exits;

[0018] If the application configuration file is parsed successfully, an application monitoring process is created and run. The application monitoring process obtains the value of the corresponding performance indicator according to the performance indicator acquisition path, and compares the value of the performance indicator with the corresponding threshold. If the performance indicator value is greater than the corresponding threshold, the performance indicator is abnormal, and the value of the abnormal performance indicator and the corresponding performance indicator name and system time are recorded.

[0019] Furthermore, the key performance indicators of the full-link tracking system include the following steps:

[0020] Parse the system monitoring configuration file and obtain the system items that enable monitoring;

[0021] For each system item that is enabled for monitoring, a corresponding system indicator monitoring process is created and run. The system indicator monitoring process parses the configuration file of the system item. After the analysis is completed, the system item is monitored and the system indicator monitoring results are obtained. If there are any abnormalities in the system indicator monitoring results, the abnormal system indicator monitoring results and the corresponding system time are recorded.

[0022] Furthermore, when recording the corresponding full system abnormal key performance indicators, it includes:

[0023] Obtain the moment when the performance indicator value of the target application is abnormal and establish the corresponding time interval;

[0024] Select all abnormal key performance indicator monitoring results within the time interval.

[0025] Furthermore, the time interval is specifically a time interval centered on the moment when the abnormal performance indicator value of the target application is monitored, the starting moment of the time interval is the difference between the moment when the abnormal performance indicator value of the target application is monitored and a preset time difference, and the ending moment of the time interval is the sum of the moment when the abnormal performance indicator value of the target application is monitored and the preset time difference.

[0026] Furthermore, the performance indicators specifically refer to user-specified performance indicators, including one or more of response time, throughput, resource utilization, and error rate, and the system indicators include one or more of CPU, memory, disk, network, and process.

[0027] Furthermore, when performing intelligent analysis and prediction, performing intelligent analysis and prediction on the full amount of performance indicator data specifically refers to using AI technology to preprocess and extract features of the collected full amount of performance indicator data to obtain feature data, building a performance prediction and anomaly detection model by training a machine learning model, inputting the feature data into the performance prediction and anomaly detection model to obtain model prediction results, and predicting potential performance problems based on the model prediction results.

[0028] Furthermore, during the intelligent tuning, resource allocation is automatically adjusted based on the predicted results of the performance indicator data and combined with expert experience. Specifically, resource allocation rules and priority constraints are defined by introducing expert experience, the current application load and predicted performance indicator data are analyzed, and real-time adjustment of the resource allocation strategy is achieved through a dynamic optimization model.

[0029] The present invention also provides an application performance monitoring system for a Linux operating system, comprising:

[0030] Application performance monitoring module, used to monitor the performance indicators of target applications in real time;

[0031] System monitoring module, used for continuous monitoring and full-link tracking of system indicators;

[0032] The application performance anomaly analysis module is used to establish an early warning mechanism. When the performance indicators of the target application are monitored to be abnormal, an alarm is triggered immediately, the abnormal indicators of the entire system are recorded, the alarm log is recorded, the corresponding abnormal system indicator monitoring results are searched, and the abnormal performance indicators of the target application and the corresponding abnormal system indicator monitoring results are recorded in the application performance anomaly log;

[0033] The intelligent analysis and prediction module uses AI technology to preprocess and extract features from the full amount of performance indicator data collected to obtain feature data. It builds a performance prediction and anomaly detection model by training the machine learning model, inputs the feature data into the performance prediction and anomaly detection model to obtain the model prediction results, and predicts potential performance problems based on the model prediction results.

[0034] The intelligent tuning module is used to introduce expert experience to define resource allocation rules and priority constraints, formulate resource optimization strategies based on application load and performance indicator data, perform resource adjustment operations through automated scripts or API interfaces, monitor performance indicator changes after resource optimization, evaluate optimization effects, and continuously iterate optimization.

[0035] The present invention also proposes a computer system, comprising a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of any one of the methods for monitoring application performance of a Linux operating system.

[0036] The present invention also provides a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of any one of the above-mentioned methods for monitoring application performance of a Linux operating system are implemented.

[0037] The present invention also provides a computer program product, comprising a computer program / instruction, which implements the steps of any one of the above-mentioned methods for monitoring application performance of a Linux operating system when executed by a processor.

[0038] Compared with the prior art, the advantages of the present invention are:

[0039] The present invention only monitors the performance indicator value of the specified application. If the application performance indicator value exceeds the specified threshold, it is abnormal. The application performance monitoring result is obtained simply and efficiently without relying on the logs of each component in the software stack.

[0040] When the present invention monitors an abnormal application performance indicator, it obtains the corresponding full system indicator monitoring results based on the abnormal time point, and saves the application performance monitoring abnormal results and the corresponding system indicator monitoring results as abnormal logs, thereby greatly reducing the number of logs and facilitating users to quickly track and analyze the causes of the abnormalities.

[0041] The present invention makes full use of AI technology and machine learning algorithms in intelligent analysis and prediction, preprocesses and extracts features from the collected performance data, builds performance prediction and anomaly detection models, and predicts potential performance problems based on the model prediction results.

[0042] During intelligent tuning, the present invention combines expert experience, dynamically adjusts resource allocation according to application load and performance indicator data, monitors changes in performance indicators after resource optimization, evaluates optimization effects, and continuously iterates optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 Schematic diagram of a flow chart of an embodiment of the present invention. DETAILED DESCRIPTION

[0044] The present invention is further described below in conjunction with the accompanying drawings and specific preferred embodiments, but the protection scope of the present invention is not limited thereby.

[0045] Before introducing the specific embodiments of the present invention, related concepts or terms are first explained.

[0046] Application performance: Application performance is the ability to evaluate whether an application can efficiently and stably process user requests and provide a good user experience during operation. Common application performance indicators include response time, throughput, resource utilization, error rate, etc.

[0047] Monitoring: Application performance monitoring systematically collects and analyzes core performance indicators during the operation of the application, such as response time, throughput, concurrent user load, and resource utilization. Through real-time and accurate monitoring mechanisms, any abnormal indicator values ​​that deviate from the normal range can be quickly identified. Once a performance anomaly is found, the monitoring tool will record application indicators and system resource anomaly information in real time, including but not limited to abnormal time, CPU, memory, network, disk IO latency, etc., to provide users with strong data support and clear troubleshooting paths, thereby accelerating problem solving and ensuring that the application continues to run efficiently and stably.

[0048] Embodiment 1

[0049] In order to solve the problems existing in the prior art, this embodiment proposes an application performance monitoring method for a Linux operating system, which realizes application performance monitoring function, system monitoring function and application performance abnormality analysis function, and can help system administrators quickly locate application performance problems and handle them. Among them:

[0050] The application performance monitoring function mainly analyzes user-configurable files, monitors the application performance indicators set by users in real time, and discovers and records abnormal application performance indicators;

[0051] The system monitoring function monitors the enabled system indicators according to the monitoring switch options in the configuration file, including system resources (CPU, memory), processes, network, disk and other information;

[0052] When the application performance anomaly analysis function monitors an abnormal application performance indicator, it reads the system indicator monitoring results in real time, and filters out the abnormal system indicator monitoring results within the time difference according to the time difference set in the configuration file and the timestamp of the current abnormal application performance indicator as the benchmark. It combines the results of the two and records the final application performance anomaly information in the application performance anomaly log.

[0053] Based on the above functions, such as Figure 1 As shown, the method of this embodiment includes the following steps:

[0054] S101) Continuously monitor the performance indicators of the target application and track the key performance indicators of the system throughout the entire link, including network throughput, CPU utilization, memory usage, disk IO, etc., and monitor the entire life cycle of business data;

[0055] S102) Establishing an early warning mechanism, if the performance indicators of the target application are monitored to be abnormal, immediately triggering an alarm, recording the corresponding full-scale abnormal system key performance indicator monitoring results, recording an alarm log, and adding the abnormal performance indicators of the target application and the corresponding full-scale abnormal key performance indicators to the alarm log;

[0056] S103) Intelligent analysis and prediction, using AI and machine learning algorithms to perform intelligent analysis and prediction on the full amount of performance indicator data to obtain the prediction results of the performance indicator data;

[0057] S104) Intelligent tuning: Based on the predicted results of the performance indicator data, combined with expert experience, resource allocation is automatically adjusted to achieve performance optimization.

[0058] The following takes the application of the method of this embodiment to the Kylin Advanced Server Operating System as an example to specifically describe each step:

[0059] In step S101 of this embodiment, the performance indicator specifically refers to the performance indicator specified by the user, including one or more of response time, throughput, resource utilization, and error rate. When the performance indicator of the target application is monitored in real time, the user-configurable file is mainly parsed, the application performance indicator set by the user is monitored in real time, and application performance anomalies are discovered and recorded. Specifically, the application indicator monitoring work is completed by parsing the application monitoring configuration file, identifying the monitoring indicator name, the monitoring indicator acquisition path, and the indicator abnormality threshold, etc., including the following steps:

[0060] S201) performing command parameter parsing, log module initialization and monitoring item initialization, parsing the application monitoring configuration file, obtaining the performance indicator name, performance indicator acquisition path and corresponding threshold of the target application;

[0061] S202) If the parsing of the application configuration file fails, then end and exit;

[0062] S203) If the application configuration file is parsed successfully, an application monitoring process is created and run. The application monitoring process obtains the value of the corresponding performance indicator according to the performance indicator acquisition path, and compares the value of the performance indicator with the corresponding threshold. If the performance indicator value is greater than the corresponding threshold, the performance indicator is abnormal. The abnormal performance indicator monitored by the application monitoring process is obtained, and the value of the abnormal performance indicator and the corresponding performance indicator name and system time are recorded.

[0063] The above steps are implemented by adding application performance indicator monitoring code and starting the service through the service. Before starting the service, the application monitoring configuration file should be modified and saved to make its configuration items compliant. After that, it is completely automatically identified, matched and implemented by the program.

[0064] In step S101 of this embodiment, the key performance indicators of the system include one or more of CPU utilization, memory usage, disk IO, network throughput, and process. When the key performance indicators of the full-link tracking system are tracked, the resources of the CPU, memory, disk, network, and process information are mainly monitored, and abnormal system monitoring data is recorded in the log, specifically, the resources of the CPU utilization, memory utilization, disk utilization, network (bandwidth and utilization), and process information are monitored, and abnormal information is recorded, including the following steps:

[0065] S301) parse command parameters, initialize log modules and monitoring items, parse system monitoring configuration files, and obtain system items for monitoring, including CPU, memory, disk, network, and process;

[0066] S302) For each system item that is enabled for monitoring, a corresponding system indicator monitoring process is created and run. The system indicator monitoring process calls the configuration parsing function to parse the configuration file of the system item. After the parsing is completed, the system item is monitored and the system indicator monitoring results (i.e., CPU usage, memory usage, disk usage, network (bandwidth and usage), process information) are obtained. If the system indicator monitoring results are abnormal (such as greater than the corresponding abnormal threshold), the abnormal system indicator monitoring results and the corresponding system time are recorded in the system log.

[0067] The above steps are implemented by adding system monitoring code, such as Figure 1 As shown, the above steps are executed after the application configuration file is parsed successfully. Before the system indicator monitoring is started, each system monitoring configuration file should be modified and saved to make its configuration items compliant. After that, the process is completely automatically identified, matched and implemented by the program.

[0068] In step S102 of this embodiment, an early warning mechanism is established. When an abnormality is detected in the application performance indicator monitoring, the application performance abnormality is analyzed, the system key performance indicator monitoring results are read in real time, and according to the time difference set in the configuration file, the abnormal system key performance indicator monitoring results within the time difference are filtered out based on the timestamp of the current abnormal application performance indicator. The results of the two are combined to record the final application performance abnormality information in the application performance abnormality log, including the following steps:

[0069] S401) respectively obtain the abnormal results of application performance indicator monitoring and system key performance indicator monitoring as inputs of the abnormal analysis module. In this embodiment, the abnormal analysis module is used to find the abnormal system key performance indicator monitoring result corresponding to the abnormal performance indicator. When finding the corresponding abnormal system key performance indicator monitoring result, first obtain the time when the performance indicator value of the target application is abnormal, and establish the corresponding time interval; then select all abnormal system key performance indicator monitoring results within the time interval;

[0070] In this embodiment, the abnormal results of application performance indicator monitoring and system key performance indicator monitoring are obtained by obtaining the value of the current abnormal performance indicator and the corresponding performance indicator name and system time from the application monitoring process, and obtaining all abnormal system key performance indicator monitoring results and the corresponding system time from the system key performance indicator monitoring process;

[0071] S402) After the abnormality analysis module obtains the time when the abnormal performance indicator value of the target application is monitored, a corresponding time interval is established. In this embodiment, the time interval is specifically a time interval centered on the time when the abnormal performance indicator value of the target application is monitored. The starting time of the time interval is the difference between the time when the abnormal performance indicator value of the target application is monitored and the preset time difference, and the ending time of the time interval is the sum of the time when the abnormal performance indicator value of the target application is monitored and the preset time difference;

[0072] S403) The abnormal analysis module selects the abnormal system key performance indicator monitoring results whose system time is within the time interval, thereby completing the screening of the abnormal system indicator monitoring results, and records the current abnormal performance indicator value and the corresponding performance indicator name and system time as well as the screened abnormal system key performance indicator monitoring results and the corresponding system time in the application performance anomaly log.

[0073] The above steps are implemented by adding application performance exception analysis code.

[0074] In step S103 of this embodiment, when the full amount of performance indicator data is intelligently analyzed and predicted, the memory network is used to model the time series characteristics of the performance data, predict future performance trends, and the neural network (CNN) is used to extract multiple features in the performance data, and possible performance bottlenecks and problem points are analyzed based on the model prediction results. Specifically, the full amount of performance indicator data is preprocessed and feature extracted by AI technology to obtain feature data, and a performance prediction and anomaly detection model is constructed by training a machine learning model, and the feature data is input into the performance prediction and anomaly detection model to obtain the model prediction results, including the following steps:

[0075] S501) Preprocessing and feature extraction of performance data: In the data preprocessing stage, a large amount of performance data is constructed in a variety of different usage scenarios. These scenarios include but are not limited to high load, low load, normal load, burst traffic, network delay, etc. By simulating and recording the system performance data in these scenarios, a rich data source is provided for subsequent model training. Dynamic noise filtering technology is used to effectively identify and eliminate high-frequency noise or abnormal data for the constructed data. By analyzing the frequency characteristics and statistical characteristics of the data, the threshold is automatically set, and the data exceeding the threshold is regarded as noise and eliminated. At the same time, a hybrid strategy based on time series model and interpolation method is combined to perform adaptive missing value completion. First, the time series model is used to predict the approximate range of missing values. According to the distribution characteristics of the data and the correlation between adjacent values, the interpolation method is used for accurate completion to improve data quality. In addition, multimodal data fusion technology is introduced for the full amount of performance indicator data to fuse data from different sources and in different formats (such as CPU usage, memory occupancy, disk I / O, network bandwidth, etc.), and the data format is unified through the embedding layer alignment. During the data segmentation process, stratified sampling is combined with a dynamic segmentation mechanism based on key performance events. According to the distribution characteristics of the data and the occurrence of key performance events (such as system crashes, performance bottlenecks, etc.), the data is divided into different levels or categories. In each level or category, the stratified sampling method is used to ensure the consistency and representativeness of the data distribution of the training set and the test set.

[0076] In the feature extraction stage, more comprehensive feature information is obtained through multi-level feature extraction such as statistical feature extraction, time series feature extraction, and high-dimensional features extracted by deep learning. Statistical feature extraction extracts a variety of statistical features from the original data, including mean, standard deviation, maximum, minimum, median, quartiles, etc., which can reflect the overall distribution and discreteness of the data and help the model capture the stability of system performance; time series feature extraction extracts a variety of time series features from the original data, such as trend features, periodic features, autocorrelation features, etc., which can reflect the change rules and periodic changes of data over time, and help the model predict the future trend of system performance; in order to capture more complex and abstract feature information, deep learning technology (neural network CNN) is used to extract high-dimensional features from the original data. By training the deep learning model, useful feature representations are automatically learned and extracted from the original data, which improves the prediction performance and generalization ability of the model.

[0077] S502) Build performance prediction and anomaly analysis models: Select different machine learning models according to specific application scenarios and data characteristics. Use multi-model fusion technology to build models. Combine different types of models such as random forests and time series models through ensemble learning to give full play to their advantages in capturing nonlinear relationships and time series modeling. Optimize cross-architecture hyperparameters for different hardware architectures (such as ARM and x86) to improve adaptability and performance. Build a multi-indicator evaluation system and dynamically adjust indicator weights in combination with Meta-Learning technology to make model evaluation more in line with application scenario requirements.

[0078] S503) Based on the trained model, real-time performance prediction and anomaly analysis are performed on the newly monitored performance data, and the model is embedded in the stream processing framework to achieve rapid discovery and response to performance issues. For the predicted abnormal data, it is classified and prioritized through a clustering algorithm to help users quickly locate the performance issues that need to be optimized most. The model interpretation tool is used to quantify the contribution of key performance indicators to the prediction results, and a hybrid method based on rules and machine learning is used to generate the analysis results of this performance data. In addition, the optimized performance data is re-injected into the training pipeline through a feedback learning mechanism, and the model threshold is dynamically adjusted in combination with reinforcement learning to build a closed-loop optimization process to continuously improve the model's prediction and analysis capabilities.

[0079] The above steps are achieved by adding intelligent analysis and prediction codes.

[0080] In step S104 of this embodiment, when automatically adjusting resource allocation according to the predicted results of performance indicator data and expert experience, the resource allocation rules and priority constraints are defined by introducing expert experience, the current application load and predicted performance indicator data are analyzed, and the real-time adjustment of resource allocation strategy is realized through dynamic optimization model. In intelligent automatic tuning, the expert experience module integrates advanced performance diagnosis logic, which can perform in-depth comprehensive analysis on the identified abnormal indicators based on the current application load and predicted performance indicator data. The potential correlation between multiple abnormal factors is comprehensively considered to comprehensively evaluate the real situation of system performance. At the same time, the system can automatically retrieve the tuning strategy library, match the most appropriate recommended index for various performance conditions, and organize these indexes into a clear list. Based on these indexes, the system extracts specific tuning measures and detailed operation steps from the configuration file to provide users with both scientific and feasible performance optimization solutions, including specific indicators of performance bottlenecks, predicted data of load, results of abnormal detection, priority sorting of tasks and specific tuning suggestions. If the user allows, resource allocation will be dynamically adjusted, such as increasing CPU and memory resources, adjusting virtual machine configuration, etc. At the same time, the system will continue to monitor changes in performance indicators after resource optimization, evaluate the optimization effect through application performance testing, and continuously improve and perfect it based on actual feedback.

[0081] When diagnosing performance based on expert experience, the performance indicators of the application are first judged. Specifically, the current application performance indicators are compared with the corresponding indicator thresholds and the current performance indicators are judged based on the comparison results to determine whether they are abnormal. If they are abnormal indicators, the system indicators (for example, CPU usage, memory usage) are judged to be abnormal. The system indicator is judged by comparing the current system indicator with the corresponding indicator thresholds and judging whether the current system indicator is abnormal based on the comparison results. If several abnormal indicators exist at the same time, the corresponding performance status and tuning suggestions are derived based on the performance diagnosis logic. The performance diagnosis logic is derived from actual tuning experience and is embedded in the program code. The following steps are included:

[0082] S601) For the performance indicators that need to be optimized, interpret the data analysis results using the expert knowledge base;

[0083] S602) Adjust the performance optimization strategy according to the expert's suggestions and formulate a specific resource adjustment plan, which should include the specific operations of resource allocation, execution time and expected results, etc.;

[0084] S603) Execute the above solution within the appropriate time window, complete the resource adjustment operation, monitor the system status during the resource adjustment process, and ensure that the adjustment operation proceeds smoothly and achieves the expected effect;

[0085] S604) Collect and process monitoring data, analyze changes in performance indicators after resource optimization, compare performance indicators before and after optimization, evaluate the effect of resource optimization, and if the optimization effect is not ideal, re-analyze the data and formulate a new tuning strategy.

[0086] The above steps are implemented by adding intelligent tuning codes.

[0087] Embodiment 2

[0088] This embodiment provides an application performance monitoring system for a Linux operating system, including:

[0089] Application performance monitoring module, used to continuously monitor the performance indicators of target applications;

[0090] System monitoring module, used for continuous monitoring and full-link tracking of system indicators;

[0091] The application performance anomaly analysis module is used to establish an early warning mechanism. When the performance indicators of the target application are monitored to be abnormal, an alarm is triggered immediately, and the corresponding abnormal key performance indicators of the entire system are recorded, and an alarm log is recorded. The abnormal performance indicators of the target application and the corresponding abnormal key performance indicators of the entire system are recorded in the application performance anomaly log;

[0092] The intelligent analysis and prediction module uses AI technology to preprocess and extract features from the full amount of performance indicator data collected to obtain feature data, which mainly includes preprocessing steps such as data cleaning, alignment, and standardization, as well as multi-level feature extraction such as statistical features, time series features, and high-dimensional features extracted by deep learning. Through noise reduction, frequency domain analysis, and machine learning models, it mines the implicit patterns and laws in the data to provide high-quality input data for performance prediction, anomaly detection, and system tuning, while improving the accuracy of the model and the pertinence of the tuning. By training the machine learning model, the performance prediction and anomaly detection model is constructed, and the feature data is input into the performance prediction and anomaly detection model to obtain the model prediction results, and potential performance problems are predicted based on the model prediction results.

[0093] The intelligent tuning module is used to introduce expert experience to define resource allocation rules and priority constraints, formulate resource optimization strategies based on application load and performance indicator data, perform resource adjustment operations through automated scripts or API interfaces, monitor performance indicator changes after resource optimization, evaluate optimization effects, and continuously iterate optimization.

[0094] This embodiment further proposes a computer system, including a memory, a processor and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the application performance monitoring method of the Linux operating system described in the first embodiment.

[0095] This embodiment further provides a computer-readable storage medium on which a computer program / instruction is stored. When the computer program / instruction is executed by a processor, the steps of the application performance monitoring method of the Linux operating system described in the first embodiment are implemented.

[0096] This embodiment further provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the application performance monitoring method of the Linux operating system described in the first embodiment.

[0097] To sum up, the present invention proposes an application performance monitoring method for a Linux operating system, and also proposes an application performance monitoring system for a Linux operating system that can implement the method of the present invention. Through innovative means such as precise monitoring of application performance indicator values, filtering and dumping of abnormal system indicators, intelligent analysis and prediction, and intelligent tuning, application performance indicators can be monitored more simply, comprehensively, accurately, and intelligently on the Kylin advanced server operating system.

[0098] The solution proposed by the present invention reduces the complexity of application performance monitoring, while being able to continuously monitor the application performance indicator status and system resource status, and can provide relatively comprehensive monitoring information when application performance is abnormal, with high availability and the following advantages:

[0099] 1. Comprehensive monitoring of application performance, including application performance monitoring, system monitoring, application performance anomaly analysis, intelligent analysis and prediction, and intelligent tuning;

[0100] 2. The tool has been run on the Kylin Advanced Server Operating System platform, and all functions provided by the tool are running normally.

[0101] The above is only a preferred embodiment of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should also be regarded as the protection scope of the present invention.

Claims

1. A method for monitoring application performance of a Linux operating system, characterized in that: The following steps are involved: Continuously monitor the performance indicators of the target application, track the key performance indicators of the system throughout the entire link, and monitor the entire life cycle of business data; If the performance indicators of the target application are monitored to be abnormal, an alarm is triggered immediately, and the corresponding abnormal key performance indicators of the entire system are recorded. The abnormal performance indicators of the target application and the corresponding abnormal key performance indicators of the entire system are added to the alarm log; Perform intelligent analysis and prediction on all performance indicator data to obtain the prediction results of the performance indicator data; Based on the prediction results of performance indicator data and combined with expert experience, resource allocation is automatically adjusted to achieve performance optimization.

2. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: When continuously monitoring the performance indicators of the target application, the following steps are included: Parse the application monitoring configuration file to obtain the performance indicator name, performance indicator acquisition path and corresponding threshold of the target application; If the parsing of the application configuration file fails, the process ends and exits; If the application configuration file is parsed successfully, an application monitoring process is created and run. The application monitoring process obtains the value of the corresponding performance indicator according to the performance indicator acquisition path, and compares the value of the performance indicator with the corresponding threshold. If the performance indicator value is greater than the corresponding threshold, the performance indicator is abnormal, and the value of the abnormal performance indicator and the corresponding performance indicator name and system time are recorded.

3. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: The key performance indicators of the full-link tracking system include the following steps: Parse the system monitoring configuration file and obtain the system items that enable monitoring; For each system item that is enabled for monitoring, a corresponding system indicator monitoring process is created and run. The system indicator monitoring process parses the configuration file of the system item. After the analysis is completed, the system item is monitored and the system indicator monitoring results are obtained. If there are any abnormalities in the system indicator monitoring results, the abnormal system indicator monitoring results and the corresponding system time are recorded.

4. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: When recording the corresponding full system abnormal key performance indicators, it includes: Obtain the moment when the performance indicator value of the target application is abnormal and establish the corresponding time interval; Select all abnormal key performance indicator monitoring results within the time interval.

5. The method for monitoring application performance of a Linux operating system according to claim 4, characterized in that: The time interval is specifically a time interval centered on the moment when the abnormal performance indicator value of the target application is monitored, the starting moment of the time interval is the difference between the moment when the abnormal performance indicator value of the target application is monitored and a preset time difference, and the ending moment of the time interval is the sum of the moment when the abnormal performance indicator value of the target application is monitored and the preset time difference.

6. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: The performance indicators specifically refer to performance indicators specified by the user, including one or more of response time, throughput, resource utilization, and error rate. The key performance indicators include one or more of CPU utilization, memory usage, disk IO, network throughput, and process.

7. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: When performing intelligent analysis and prediction on the full amount of performance indicator data, the full amount of performance indicator data is preprocessed and feature extracted to obtain feature data, a performance prediction and anomaly detection model is built by training a machine learning model, and the feature data is input into the performance prediction and anomaly detection model to obtain the model prediction result.

8. The method for monitoring application performance of a Linux operating system according to claim 1, characterized in that: When automatically adjusting resource allocation based on the predicted results of performance indicator data and combined with expert experience, the resource allocation rules and priority constraints are defined by introducing expert experience, the current application load and predicted performance indicator data are analyzed, and real-time adjustment of resource allocation strategies is achieved through a dynamic optimization model.

9. An application performance monitoring system for a Linux operating system, characterized in that: include: Application performance monitoring module, used to monitor the performance indicators of target applications in real time; System monitoring module, used to continuously monitor and track the key performance indicators of the system throughout the entire link; The application performance anomaly analysis module is used to establish an early warning mechanism. When the performance indicators of the target application are monitored to be abnormal, an alarm is triggered immediately, and the corresponding key performance indicators of the full system abnormality are recorded and added to the alarm log; The intelligent analysis and prediction module uses AI technology to preprocess and extract features from all performance indicator data to obtain feature data. It builds performance prediction and anomaly detection models by training machine learning models, inputs feature data into the performance prediction and anomaly detection models to obtain model prediction results, and predicts potential performance issues based on the model prediction results. The intelligent tuning module is used to introduce expert experience to define resource allocation rules and priority constraints, analyze the current application load and predicted performance indicator data, formulate resource optimization strategies, perform resource adjustment operations through automated scripts or API interfaces, monitor performance indicator changes after resource optimization, evaluate optimization effects, and continuously iterate optimization.

10. A computer system comprising a memory, a processor and a computer program stored in the memory, characterized in that: The processor executes the computer program to implement the steps of the application performance monitoring method of the Linux operating system according to any one of claims 1 to 8.

Images