An intelligent analysis method for virtual machine status

Abstract

The present invention provides an intelligent analysis method for virtual machine states, belonging to the technical field of data processing, including: acquiring and collecting real-time performance data and real-time log data of a virtual machine to determine the real-time operation data of the virtual machine; obtaining the historical operation data of the virtual machine, determining a type prediction model for multiple error type tags based on the historical operation data, and determining the prediction model; analyzing the real-time operation data of the virtual machine and the prediction model to determine the real-time operation state of the virtual machine; determining whether to send an alarm message based on the real-time operation state of the virtual machine and generating a status report. A refined prediction model can be constructed to achieve accurate personalized state prediction and early warning, improve the automation degree and response speed of virtual machine management, optimize the operation and maintenance process, reduce the possibility of human intervention and errors, can improve the efficiency of abnormal state handling, enhance the stability and reliability of the virtual machine, and improve the automation level of virtual machine management.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular, to an intelligent analysis method for virtual machine states. Background Art

[0002] Since the virtual machine technology was first proposed in the late 1990s, it has become a widely used virtualization technology in modern computing environments. By creating multiple virtual instances on physical hardware, virtual machines enable different operating systems to run independently on the same hardware, thereby improving the utilization rate of hardware resources and the flexibility of management. In the early days, the monitoring of the running state of virtual machines mainly relied on traditional performance monitoring tools, which usually only provided basic monitoring of resource usage and were difficult to capture complex problems in virtual machines in real time. However, with the in-depth application of virtualization technology, how to efficiently and real-time manage the state and performance of virtual machines has become a challenge.

[0003] Therefore, the present invention provides an intelligent analysis method for virtual machine states. Summary of the Invention

[0004] The present invention provides an intelligent analysis method for virtual machine states, which constructs a prediction model according to the obtained historical operation data, analyzes the obtained real-time operation data and the prediction model, determines the real-time operation state of the virtual machine, determines whether to send an alarm message based on the real-time operation state of the virtual machine, and generates a status report. A refined prediction model can be constructed to achieve accurate personalized state prediction and early warning, improve the automation degree and response speed of virtual machine management, optimize the operation and maintenance process, reduce the possibility of human intervention and errors, improve the efficiency of abnormal state handling, improve the stability and reliability of virtual machines, and improve the automation level of virtual machine management.

[0005] The present invention provides an intelligent analysis method for virtual machine states, including:

[0006] 101: Obtain and collect the real-time performance data and real-time log data of the virtual machine to determine the real-time operation data of the virtual machine;

[0007] 102: Obtain the historical operation data of the virtual machine, determine the type prediction models of multiple error type labels based on the historical operation data, and determine the prediction model;

[0008] 103: Analyze the real-time operation data of the virtual machine and the prediction model to determine the real-time operation state of the virtual machine;

[0009] 104: Determine whether to send an alarm message based on the real-time operation state of the virtual machine, and generate a status report.

[0010] A method for intelligent analysis of virtual machine status provided by the present invention collects real-time performance data and real-time log data of a virtual machine in real time to determine real-time operation data of the virtual machine, including:

[0011] Based on the API interface of the virtual machine, obtain performance metrics of multiple performances of the virtual machine within a real-time specified time period, and determine real-time performance data based on the performance metrics of all performances of the virtual machine;

[0012] Obtain real-time log data of the virtual machine within a real-time specified time period in real time, where the real-time log data includes real-time operation logs and real-time error logs;

[0013] Preprocess the real-time performance data and the real-time log data respectively to determine the real-time operation data of the virtual machine.

[0014] A method for intelligent analysis of virtual machine status provided by the present invention obtains historical operation data of a virtual machine, determines a type prediction model for multiple error type tags based on the historical operation data, and determines a prediction model, including:

[0015] Obtain historical operation sub-data of the virtual machine within multiple historical specified time periods, where the historical operation sub-data includes historical performance data, historical log data, and historical status data, and the historical log data includes historical operation logs and historical error logs;

[0016] Extract historical operation sub-data with historical error logs in the historical log data, and determine historical operation data based on the extracted historical operation sub-data within all historical time periods;

[0017] Analyze the historical error logs in the historical log data of the historical operation sub-data in each historical time period in the historical operation data, and determine the error log tags for each historical error log, where the error log tags include error type sub-tags, error severity sub-tags, error cause sub-tags, and error impact range sub-tags;

[0018] Classify all historical operation sub-data in the historical operation data based on the error type sub-tags and error severity sub-tags in the error log tags of all historical error logs in the historical operation data, and determine type historical data and error type tags of the type historical data, where the type historical data includes multiple historical operation sub-data with the same error type sub-tags and the same error severity sub-tags;

[0019] Train a type prediction model for each error type tag based on each type of historical data, and determine a prediction model based on all type prediction models.

[0020] A method for intelligent analysis of virtual machine states provided by the present invention trains a type prediction model for each error type label based on historical data of each type, including:

[0021] Extract features from the historical performance data of the historical operation sub-data within each historical specified time period in the type historical data to determine the performance feature vector of the historical performance data of the historical operation sub-data within each historical specified time period in the type historical data;

[0022] Extract features from the historical operation logs in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data to determine the operation feature vector of the historical operation logs in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data;

[0023] Based on the error log labels, performance feature vectors, and operation feature vectors of the historical operation sub-data within all historical specified time periods in the type historical data, determine the operation-performance correlation value of each type of historical data;

[0024] Determine the error log labels, performance feature vectors, operation feature vectors, and operation-performance correlation values of the historical operation sub-data within all historical specified time periods in each type of historical data as the input of the type prediction model for each error type label, and determine the historical state data of the historical operation sub-data within all historical specified time periods in each type of historical data as the output of the type prediction model for each error type label. Train the type prediction model for each error type label based on the historical operation sub-data within all historical specified time periods in each type of historical data.

[0025] A method for intelligent analysis of virtual machine states provided by the present invention determines the operation-performance correlation value of each type of historical data based on the error log labels, performance feature vectors, and operation feature vectors of the historical operation sub-data within all historical specified time periods in the type historical data, including:

[0026] Among them, represents the operation-performance correlation value of the f-th type of historical data, respectively represent the first sub-operation-performance correlation value and the second sub-operation-performance correlation value of the historical operation sub-data in the t-th historical specified time period in the f-th type of historical data, and a1 and a2 respectively represent the weights of the first sub-operation-performance correlation value and the second sub-operation-performance correlation value, represents the period decay factor of the f-th type of historical data, represents the number of historical operation sub-data in the f-th type of historical data, Denote the eigenvalue of the \(i\)-th operation feature of the operation feature vector of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data. Denote the eigenvalue of the \(j\)-th performance feature of the performance feature vector of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data. Denote the number of operation features in the operation feature vector. Denote the number of performance features in the performance feature vector. Denote the interaction influence coefficient of the \(i\)-th operation feature of the operation feature vector and the \(j\)-th performance feature of the performance feature vector of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data. Denote the influence value of the error cause sub-label of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data on the operation feature vector. Denote the influence value of the error influence range sub-label of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data on the operation feature vector. Denote the first correlation coefficient between the error cause sub-label and the \(i\)-th operation feature in the operation feature vector. Denote the second correlation coefficient between the error influence range sub-label and the \(i\)-th operation feature in the operation feature vector. Denote the influence value of the error type sub-label and the error severity sub-label of the type of historical data on the operation feature vector. Denote the weight of the \(i\)-th operation feature of the operation feature vector of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data. Denote the adjustment factor in the \(t\)-th historical specified time period in the \(f\)-th type of historical data. Denote the weight of the \(j\)-th performance feature of the performance feature vector of the historical operation sub-data in the \(t\)-th historical specified time period in the \(f\)-th type of historical data.

[0027] According to an intelligent virtual machine state analysis method provided by the present invention, analyze the real-time operation data of the virtual machine and the prediction model to determine the real-time operation state of the virtual machine, including:

[0028] Judge whether the real-time error log in the real-time log data in the real-time operation data of the virtual machine is empty;

[0029] If it is not empty, determine that the real-time operation state of the virtual machine is abnormal operation, and determine the abnormal operation state of the virtual machine based on the real-time operation data and the prediction model. If it is empty, determine that the real-time operation state of the virtual machine is normal operation.

[0030] A method for intelligent analysis of virtual machine status provided by the present invention determines the abnormal running status of a virtual machine based on the real-time running data of the virtual machine and a prediction model, including:

[0031] Extract features from the real-time performance data in the real-time running data to determine the real-time performance vector of the virtual machine;

[0032] Extract features from the real-time operation logs in the real-time log data in the real-time running data to determine the real-time operation vector of the virtual machine;

[0033] Analyze the real-time error logs in the real-time log data in the real-time running data to determine real-time log tags, where the real-time log tags include error type sub-tags and error severity sub-tags;

[0034] Determine the real-time error type based on the real-time log tags, and determine the type prediction model of the real-time running data based on the real-time error type;

[0035] Input the real-time performance vector and the real-time operation vector into the type prediction model of the real-time running data, and determine the abnormal running status of the virtual machine based on the output result of the type prediction model of the real-time running data.

[0036] A method for intelligent analysis of virtual machine status provided by the present invention determines the real-time running status of a virtual machine based on the real-time running data of the virtual machine and a prediction model, including:

[0037] If the real-time running status of the virtual machine is normal operation, no warning message is sent;

[0038] If the real-time running status of the virtual machine is abnormal operation, a warning message is sent based on the abnormal running status, and a status report is generated based on the real-time running data, the real-time performance vector, the real-time operation vector, the abnormal running status, and the warning message.

[0039] Compared with the prior art, the beneficial effects of the present application are as follows:

[0040] Construct a prediction model based on the obtained historical running data, analyze the obtained real-time running data and the prediction model, determine the real-time running status of the virtual machine, determine whether to send a warning message based on the real-time running status of the virtual machine, and generate a status report. A refined prediction model can be constructed to achieve accurate personalized status prediction and warning, improve the automation level and response speed of virtual machine management, optimize the operation and maintenance process, reduce the possibility of human intervention and errors, can improve the efficiency of abnormal status handling, improve the stability and reliability of the virtual machine, and improve the automation level of virtual machine management. Description of the Drawings

[0041] To more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the accompanying drawings required for the description of the embodiments or the prior art. Obviously, the accompanying drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other accompanying drawings can be obtained based on these drawings.

[0042] Figure 1 It is a schematic flowchart of a method for intelligent analysis of virtual machine status provided by an embodiment of the present invention. Detailed implementation manners

[0043] To make the objectives, technical solutions, and advantages of the present invention clearer, the following will clearly and completely describe the technical solutions in the present invention with reference to the accompanying drawings in the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts belong to the scope of protection of the present invention.

[0044] Embodiment 1:

[0045] An embodiment of the present invention provides a method for intelligent analysis of virtual machine status. As Figure 1 shown, it includes:

[0046] 101: Obtain and collect the real-time performance data and real-time log data of the virtual machine, and determine the real-time operation data of the virtual machine;

[0047] 102: Obtain the historical operation data of the virtual machine, determine the type prediction models of multiple error type tags based on the historical operation data, and determine the prediction models;

[0048] 103: Analyze the real-time operation data of the virtual machine and the prediction models, and determine the real-time operation status of the virtual machine;

[0049] 104: Determine whether to send an alarm message based on the real-time operation status of the virtual machine, and generate a status report.

[0050] In this embodiment, through the API interface of the virtual machine, the real-time performance data (such as CPU, memory, disk I / O, etc.) and log data (including real-time operation logs and real-time error logs) of the virtual machine are collected in real time.

[0051] In this embodiment, the operation data of the virtual machine in multiple historical time periods is collected, including historical performance data, historical log data, and historical status data. Based on these historical data, multiple type prediction models are constructed through machine learning.

[0052] In this embodiment, the real-time operation data of the virtual machine is analyzed by combining the real-time collected data and the established prediction model.

[0053] In this embodiment, according to the real-time operation state of the virtual machine, it is determined whether an alarm message needs to be sent. If the virtual machine is in an abnormal state, an alarm message will be immediately sent to notify relevant personnel. At the same time, a detailed status report will also be generated.

[0054] Beneficial effects of the above technical solution: A prediction model is constructed based on the obtained historical operation data, the obtained real-time operation data and the prediction model are analyzed to determine the real-time operation state of the virtual machine. Whether to send an alarm message is determined based on the real-time operation state of the virtual machine, and a status report is generated. A refined prediction model can be constructed to achieve accurate personalized status prediction and early warning, improve the automation degree and response speed of virtual machine management, optimize the operation and maintenance process, reduce the possibility of human intervention and errors, can improve the efficiency of abnormal state handling, improve the stability and reliability of the virtual machine, and improve the automation level of virtual machine management.

[0055] Embodiment 2:

[0056] An embodiment of the present invention provides a method for intelligent analysis of virtual machine status, which collects real-time performance data and real-time log data of the virtual machine in real time to determine the real-time operation data of the virtual machine, including:

[0057] Based on the API interface of the virtual machine, obtain the performance metrics of multiple performances of the virtual machine within a real-time specified time period, and determine the real-time performance data based on the performance metrics of all performances of the virtual machine;

[0058] Obtain the real-time log data of the virtual machine within a real-time specified time period in real time, where the real-time log data includes real-time operation logs and real-time error logs;

[0059] Preprocess the real-time performance data and the real-time log data respectively to determine the real-time operation data of the virtual machine.

[0060] In this embodiment, the API interface represents a program interface through which interaction with virtual machine management or other monitoring can be carried out to obtain real-time performance data and log data.

[0061] In this embodiment, through the API interface of the virtual machine, multiple performance metrics of the virtual machine within a real-time specified time period can be obtained, such as: CPU usage rate, memory occupancy, network traffic, disk I / O, etc. These performance metrics reflect the resource consumption and operation efficiency of the virtual machine at a specific moment. These performance metrics are combined together to generate real-time performance data.

[0062] In this embodiment, log data generated by the virtual machine is obtained in real time through an interface, including: real-time operation logs: recording operations performed by the virtual machine, such as startup, shutdown, configuration changes, etc.; real-time error logs: recording any errors or abnormal events occurring during the operation of the virtual machine.

[0063] In this embodiment, preprocessing is performed on the real-time performance data and real-time log data. The preprocessing steps may include denoising, data cleaning, standardization, etc. to ensure data quality and accuracy. The preprocessed data will be used to generate the final real-time operation data of the virtual machine.

[0064] Beneficial effects of the above technical solution: Real-time collection of real-time performance data and real-time log data of the virtual machine, determination of the real-time operation data of the virtual machine, can comprehensively reflect the current state of the virtual machine, and provide a data basis for determining the real-time operation state of the virtual machine.

[0065] Embodiment 3:

[0066] An embodiment of the present invention provides a method for intelligent analysis of virtual machine status, obtaining historical operation data of the virtual machine, determining a type prediction model for multiple error type labels based on the historical operation data, and determining a prediction model, including:

[0067] Obtain historical operation sub-data of the virtual machine within multiple historical specified time periods, where the historical operation sub-data includes historical performance data, historical log data, and historical status data, and the historical log data includes historical operation logs and historical error logs;

[0068] Extract historical operation sub-data with historical error logs in the historical log data, and determine historical operation data based on the extracted historical operation sub-data within all historical time periods;

[0069] Analyze the historical error logs in the historical log data of the historical operation sub-data within each historical time period in the historical operation data, and determine the error log labels for each historical error log, where the error log labels include error type sub-labels, error severity sub-labels, error cause sub-labels, and error impact scope sub-labels;

[0070] Classify all historical operation sub-data in the historical operation data based on the error type sub-labels and error severity sub-labels in the error log labels of all historical error logs in the historical operation data, and determine type historical data and the error type labels of the type historical data, where the type historical data includes multiple historical operation sub-data with the same error type sub-labels and the same error severity sub-labels;

[0071] Train the type prediction model for each error type label based on each type of historical data, and determine the prediction model based on all type prediction models.

[0072] In this embodiment, historical operation sub-data containing historical error logs is extracted from historical log data. These data reflect errors or abnormal events that occurred within a specific time period. Based on these data, historical operation data is determined.

[0073] In this embodiment, the historical error logs in each historical operation sub-data are analyzed, and error log tags are assigned to each error log. These tags include: error type sub-tag: the type of error (such as crash, performance bottleneck, etc.); error severity sub-tag: the severity of the error (such as minor, severe, etc.); error cause sub-tag: the root cause of the error (such as hardware failure, software bug, etc.); error impact scope sub-tag: the scope of influence of the error (such as local impact, global impact, etc.).

[0074] In this embodiment, according to the error type sub-tag and the error severity sub-tag, all historical operation sub-data in the historical operation data are classified. Each class of data will contain multiple historical operation sub-data, and these data have the same tags in terms of error type and error severity.

[0075] In this embodiment, the classified type historical data is used for training to construct a prediction model. Each error type tag corresponds to an independent type prediction model.

[0076] Advantages of the above technical solution: Obtain the historical operation data of the virtual machine, determine the type prediction models of multiple error type tags based on the historical operation data, and determine the prediction models, so as to construct a refined prediction model, perform personalized prediction and warning, improve the accuracy of prediction, reduce the downtime and maintenance cost of the virtual machine, and optimize the overall operation and maintenance management efficiency.

[0077] Embodiment 4:

[0078] An embodiment of the present invention provides an intelligent analysis method for virtual machine status. Based on each type of historical data, a type prediction model for each error type tag is trained, including:

[0079] Feature extraction is performed on the historical performance data of the historical operation sub-data within each historical specified time period in the type historical data to determine the performance feature vector of the historical performance data of the historical operation sub-data within each historical specified time period in the type historical data;

[0080] Feature extraction is performed on the historical operation logs in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data to determine the operation feature vector of the historical operation logs in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data;

[0081] Determine the operation - performance association value for each type of historical data based on the error log tags, performance feature vectors, and operation feature vectors of the historical operation sub - data within all historical specified time periods in the type historical data;

[0082] Determine the error log tags, performance feature vectors, operation feature vectors, and operation - performance association values of the historical operation sub - data within all historical specified time periods in each type of historical data as the input of the type prediction model for each error type label, and determine the historical status data of the historical operation sub - data within all historical specified time periods in each type of historical data as the output of the type prediction model for each error type label. Train the type prediction model for each error type label based on the historical operation sub - data within all historical specified time periods in each type of historical data.

[0083] In this embodiment, historical performance data is extracted from each historical operation sub - data, and through feature extraction, each extracted historical performance data is converted into a performance feature vector.

[0084] In this embodiment, feature extraction is performed on the historical operation logs in the historical log data, and the key information (such as operation type, time, operation object, etc.) in the historical operation logs is converted into an operation feature vector, so as to quantify the operation behavior of the virtual machine within a specific time period.

[0085] In this embodiment, by combining the error log tags, performance feature vectors, and operation feature vectors, an operation - performance association value is calculated, which reflects the relationship between the operation behavior and performance change of the virtual machine for each type of historical data.

[0086] In this embodiment, taking the error log tags, performance feature vectors, operation feature vectors, and operation - performance association values as the input, and the historical status data within each historical period as the output of the prediction model, a type prediction model for each error type label is constructed.

[0087] In this embodiment, the historical status data records the health status of the virtual machine in the past period of time.

[0088] In this embodiment, each model is specifically trained for a certain error type label to improve the accuracy of predicting the occurrence of this type of error.

[0089] The beneficial effects of the above - mentioned technical solution: Train the type prediction model for each error type label based on each type of historical data, and through the customized prediction model, achieve effective early warning for each error type, improve the prediction accuracy, and enhance the intelligent level of virtual machine management.

[0090] Example 5:

[0091] An embodiment of the present invention provides an intelligent analysis method for virtual machine states. Based on the error log tags, performance feature vectors, and operation feature vectors of historical operation sub-data within all historical specified time periods in the type historical data, the operation-performance association value of each type of historical data is determined, including:

[0092] where, represents the operation-performance association value of the f-th type of historical data, respectively represent the first sub-operation-performance association value and the second sub-operation-performance association value of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data. a1 and a2 respectively represent the weights of the first sub-operation-performance association value and the second sub-operation-performance association value, represents the period decay factor of the f-th type of historical data, represents the number of historical operation sub-data in the f-th type of historical data, represents the eigenvalue of the i-th operation feature of the operation feature vector of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data, represents the eigenvalue of the j-th performance feature of the performance feature vector of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data, represents the number of operation features in the operation feature vector, represents the number of performance features in the performance feature vector, represents the interaction influence coefficient between the i-th operation feature of the operation feature vector and the j-th performance feature of the performance feature vector of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data, represents the influence value of the error cause sub-tag of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data on the operation feature vector, represents the influence value of the error influence range sub-tag of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data on the operation feature vector, represents the first correlation coefficient between the error cause sub-tag and the i-th operation feature in the operation feature vector, represents the second correlation coefficient between the error influence range sub-tag and the i-th operation feature in the operation feature vector, represents the influence value of the error type sub-tag and the error severity sub-tag of the type historical data on the operation feature vector, represents the weight of the i-th operation feature of the operation feature vector of the historical operation sub-data in the t-th historical specified time period of the f-th type of historical data, Denote the adjustment factor for the t-th historical specified time period in the f-th type of historical data. Denote the weight of the j-th performance characteristic of the performance characteristic vector of the historical operation sub-data for the t-th historical specified time period in the f-th type of historical data.

[0093] In this embodiment, the first sub-operation - performance correlation value of the historical operation sub-data for the t-th historical specified time period in the f-th type of historical data Denote the interaction correlation value of the performance characteristic vector and the operation characteristic vector for the t-th historical specified time period in the f-th type of historical data.

[0094] In this embodiment, the second sub-operation - performance correlation value of the historical operation sub-data for the t-th historical specified time period in the f-th type of historical data Denote the interaction correlation value of the error log label and the operation characteristic vector for the t-th historical specified time period in the f-th type of historical data.

[0095] In this embodiment, the periodic decay factor of the f-th type of historical data Determined according to the number of historical operation sub-data in the f-th type of historical data and the timestamp label of the historical log data in the historical operation sub-data.

[0096] Beneficial effects of the above technical solution: Based on the error log labels, performance characteristic vectors, and operation characteristic vectors of the historical operation sub-data within all historical specified time periods in the type of historical data, determine the operation - performance correlation value for each type of historical data, which can provide a data basis for constructing a customized type prediction model for each error type label, improve the prediction accuracy, and enhance the intelligent level of virtual machine management.

[0097] Embodiment 6:

[0098] An embodiment of the present invention provides an intelligent analysis method for virtual machine status, which analyzes the real-time operation data of the virtual machine and the prediction model to determine the real-time operation status of the virtual machine, including:

[0099] Judge whether the real-time error log in the real-time log data in the real-time operation data of the virtual machine is empty;

[0100] If it is not empty, determine that the real-time operation status of the virtual machine is abnormal operation, and determine the abnormal operation status of the virtual machine based on the real-time operation data and the prediction model. If it is empty, determine that the real-time operation status of the virtual machine is normal operation.

[0101] In this embodiment, check whether the real-time error log in the real-time log data in the real-time operation data of the virtual machine is empty. The real-time error log records any errors or abnormal events generated during the real-time operation of the virtual machine.

[0102] In this embodiment, if the real-time error log is not empty, it indicates that an error or exception event has occurred in the virtual machine during the current running cycle, and it will be determined that the virtual machine is in an abnormal running state. Next, based on the real-time running data and the prediction model, the abnormal running state is further determined.

[0103] In this embodiment, if the real-time error log is empty, it means that no error or exception event has occurred in the virtual machine currently, and it is considered that the virtual machine is in a normal running state. At this time, the resource consumption, performance metrics, etc. of the virtual machine are all within the normal range, and no failure has occurred.

[0104] Beneficial effects of the above technical solution: By analyzing the real-time running data of the virtual machine and the prediction model, and determining the real-time running state of the virtual machine, potential abnormal states can be identified in real time, the accuracy and response speed of abnormal detection are improved, the maintenance and management of the virtual machine are optimized, and the stability and reliability of the virtual machine are enhanced.

[0105] Embodiment 7:

[0106] An embodiment of the present invention provides an intelligent analysis method for virtual machine status. Based on the real-time running data of the virtual machine and the prediction model, the abnormal running state of the virtual machine is determined, including:

[0107] Extract features from the real-time performance data in the real-time running data to determine the real-time performance vector of the virtual machine;

[0108] Extract features from the real-time operation logs in the real-time log data in the real-time running data to determine the real-time operation vector of the virtual machine;

[0109] Analyze the real-time error log in the real-time log data in the real-time running data to determine the real-time log tag, where the real-time log tag includes an error type sub-tag and an error severity sub-tag;

[0110] Determine the real-time error type based on the real-time log tag, and determine the type prediction model of the real-time running data based on the real-time error type;

[0111] Input the real-time performance vector and the real-time operation vector into the type prediction model of the real-time running data, and determine the abnormal running state of the virtual machine based on the output result of the type prediction model of the real-time running data.

[0112] In this embodiment, real-time performance data is extracted from the real-time running data of the virtual machine. After these performance data are subjected to feature extraction, they are transformed into performance vectors, which is a numerical set used to describe the current performance state of the virtual machine.

[0113] In this embodiment, the real-time operation logs in the real-time log data of the analysis virtual machine are analyzed, and key information therein (such as operation type, operation target, execution time, etc.) is extracted to generate an operation vector representing the operation behavior characteristics of the virtual machine.

[0114] In this embodiment, the real-time error logs are analyzed and real-time log tags are assigned to the error logs.

[0115] In this embodiment, the real-time error type is determined according to the real-time log tags, and a corresponding type prediction model is selected according to the determined real-time error type. This model is specifically used to predict the abnormal operation of the virtual machine related to the real-time error type.

[0116] In this embodiment, the generated real-time performance vector and real-time operation vector are input into the corresponding type prediction model for prediction. According to the output result of the model, the abnormal operation state of the virtual machine is determined.

[0117] The beneficial effects of the above technical solution: Based on the real-time operation data of the virtual machine and the prediction model, the abnormal operation state of the virtual machine can be determined, the real-time error type can be determined, accurate prediction can be achieved, the accuracy and response efficiency of abnormal detection can be improved, the downtime risk can be reduced, and resource management can be optimized.

[0118] Embodiment 8:

[0119] An embodiment of the present invention provides an intelligent analysis method for the virtual machine state. Based on the real-time operation data of the virtual machine and the prediction model, the real-time operation state of the virtual machine is determined, including:

[0120] If the real-time operation state of the virtual machine is normal operation, no alarm information is sent.

[0121] If the real-time operation state of the virtual machine is abnormal operation, alarm information is sent based on the abnormal operation state, and a status report is generated based on the real-time operation data, real-time performance vector, real-time operation vector, abnormal operation state, and alarm information.

[0122] In this embodiment, when the real-time operation state of the virtual machine is determined to be abnormal operation, alarm information will be sent immediately. These alarm information will be notified to relevant personnel through appropriate channels (such as emails, text messages, console notifications, etc.) to remind them that the virtual machine has an abnormal state.

[0123] In this embodiment, a status report is automatically generated based on information such as real-time operation data, real-time performance vector, real-time operation vector, and abnormal operation state.

[0124] Advantages of the above technical solution: Based on the real-time operation data of the virtual machine and the prediction model, the real-time operation state of the virtual machine is determined, which can improve the efficiency of abnormal state handling and the reliability of the virtual machine, enhance the automation level of virtual machine management, and reduce the need for manual intervention.

[0125] 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 efforts.

[0126] 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 essence of the above technical solution, or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, 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.

[0127] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention 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 described in the foregoing embodiments or perform equivalent replacements for some of the technical features. However, 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 invention.

Claims

1. A method for intelligent analysis of virtual machine status, characterized in that: include: 101: Acquire and collect real-time performance data and real-time log data of the virtual machine, and determine the real-time operation data of the virtual machine; 102: Obtain historical operation data of the virtual machine, determine a type prediction model of multiple error type labels based on the historical operation data, and determine the prediction model; 103: Analyze the real-time operation data of the virtual machine and the prediction model to determine the real-time operation status of the virtual machine; 104: Determine whether to issue an alarm message based on the real-time running status of the virtual machine, and generate a status report; Wherein, step 102 includes: Obtaining historical operation sub-data of the virtual machine within multiple historical specified time periods, wherein the historical operation sub-data includes historical performance data, historical log data, and historical status data, wherein the historical log data includes historical operation logs and historical error logs; Extracting historical operation sub-data containing historical error logs from the historical log data, and determining historical operation data based on the extracted historical operation sub-data within all historical time periods; Analyze the historical error logs in the historical log data of the historical operation sub-data in each historical time period in the historical operation data, and determine the error log label of each historical error log, wherein the error log label includes an error type sub-label, an error severity sub-label, an error cause sub-label, and an error impact range sub-label; Based on the error type subtag and the error severity subtag in the error log tags of all the historical error logs in the historical operation data, classify all the historical operation subdata in the historical operation data, determine the type historical data and the error type tag of the type historical data, wherein the type historical data includes a plurality of historical operation subdata with the same error type subtag and the same error severity subtag; Based on each type of historical data, a type prediction model for each error type label is trained, and a prediction model is determined based on all type prediction models, including: Extracting features from historical performance data of historical operation sub-data within each historical specified time period in the historical data of the type, and determining a performance feature vector of historical performance data of historical operation sub-data within each historical specified time period in the historical data of the type; Perform feature extraction on the historical operation log in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data, and determine the operation feature vector of the historical operation log in the historical log data of the historical operation sub-data within each historical specified time period in the type historical data; Determine the operation-performance correlation value of each type of historical data based on the error log labels, performance feature vectors, and operation feature vectors of all historical operation sub-data within a specified historical time period in the type of historical data; Determine the error log labels, performance feature vectors, operation feature vectors and operation-performance correlation values ​​of all historical operation sub-data within a historical specified time period in each type of historical data as the input of the type prediction model for each error type label; determine the historical status data of all historical operation sub-data within a historical specified time period in each type of historical data as the output of the type prediction model for each error type label; and train the type prediction model for each error type label based on the historical operation sub-data within all historical specified time periods in each type of historical data.

2. A method for intelligent analysis of virtual machine status according to claim 1, characterized in that: Collect real-time performance data and real-time log data of virtual machines in real time to determine the real-time operation data of virtual machines, including: Based on the API interface of the virtual machine, the performance indicators of multiple performances of the virtual machine within a real-time specified time period are obtained, and the real-time performance data is determined based on the performance indicators of all performances of the virtual machine; Acquire real-time log data of the virtual machine within a specified time period in real time, wherein the real-time log data includes real-time operation logs and real-time error logs; The real-time performance data and the real-time log data are preprocessed respectively to determine the real-time operation data of the virtual machine.

3. A method for intelligent analysis of virtual machine status according to claim 2, characterized in that: Analyze the real-time operation data of the virtual machine and the prediction model to determine the real-time operation status of the virtual machine, including: Determine whether the real-time error log in the real-time log data in the real-time running data of the virtual machine is empty; If it is not empty, determine that the real-time operating status of the virtual machine is abnormal operation, and determine the abnormal operating status of the virtual machine based on the real-time operating data and the prediction model; if it is empty, determine that the real-time operating status of the virtual machine is normal operation.

4. A method for intelligent analysis of virtual machine status according to claim 3, characterized in that: Based on the real-time operation data of the virtual machine and the prediction model, determine the abnormal operation status of the virtual machine, including: Extracting features from real-time performance data in the real-time running data to determine the real-time performance vector of the virtual machine; Extracting features from the real-time operation log in the real-time log data in the real-time operation data to determine the real-time operation vector of the virtual machine; Analyze the real-time error log in the real-time log data in the real-time operation data to determine the real-time log tag, wherein the real-time log tag includes an error type sub-tag and an error severity sub-tag; Determine the real-time error type based on the real-time log tag, and determine the type prediction model of the real-time running data based on the real-time error type; The real-time performance vector and the real-time operation vector are input into a type prediction model of the real-time operation data, and the abnormal operation state of the virtual machine is determined based on an output result of the type prediction model of the real-time operation data.

5. A method for intelligent analysis of virtual machine status according to claim 4, characterized in that: Based on the real-time operation data of the virtual machine and the prediction model, the real-time operation status of the virtual machine is determined, including: If the real-time running status of the virtual machine is normal operation, no alarm information is issued; If the real-time running status of the virtual machine is abnormal, an alarm message is issued based on the abnormal running status, and a status report is generated based on the real-time running data, the real-time performance vector, the real-time operation vector, the abnormal running status and the alarm message.

Images