Virtual machine monitoring method in cloud computing based on vector autoregression

Abstract

The invention discloses a virtual machine monitoring method in cloud computing based on vector autoregression, and particularly relates to the technical field of virtual machine data monitoring. Comprising the steps of distributed monitoring node deployment, virtual machine performance data collection, virtual machine resource demand state monitoring, virtual machine resource allocation strategy management, virtual machine performance optimization supervision, virtual machine performance optimization benefit evaluation and vector autoregression model virtual machine performance anomaly detection. According to the method, the virtual machine performance data of each data acquisition frequency of the target virtual machine monitoring node is acquired, the virtual machine resource demand overload index and the virtual machine load balanced distribution index are calculated, and the virtual machine resource optimization supervision coefficient is further obtained, so that the evaluation of a high-load virtual machine is facilitated, and the evaluation efficiency is improved. According to the method and the system, the task is distributed to other low-load virtual machines, so that the stable operation of the whole system is kept, the problem of abnormal resource distribution is timely responded by adopting a monitoring strategy, and meanwhile, the performance stability of the virtual machines is more accurately identified through vector autoregression.

Description

Technical Field

[0001] The present invention relates to the technical field of virtual machine data monitoring, and more particularly to a method for monitoring virtual machines in cloud computing based on vector autoregression. Background Art

[0002] With the rapid development of information technology, cloud computing has become an important part of modern enterprise information infrastructure. In the cloud computing environment, more and more enterprises and organizations are migrating key businesses to cloud platforms, which has led to an increasing demand for monitoring virtual machines. By monitoring their operating status, potential problems can be discovered and resolved in a timely manner, thereby ensuring service continuity and reliability.

[0003] In practical applications, the vector autoregression model provides a powerful time series analysis tool that can learn the behavior patterns of virtual machines from historical data, predict future states, and promptly identify potential performance bottlenecks and abnormal behaviors, thereby achieving optimal resource allocation and preventive fault management. The model's multivariate time series analysis capability gives it a unique advantage in processing complex and dynamically changing virtual machine monitoring data in cloud computing.

[0004] However, in actual use, it still has some shortcomings. For example, the existing virtual machine performance monitoring indicators cannot be set according to the actual usage scenarios and needs of the virtual machine. The alarm function of the monitoring tool or platform fails and cannot issue alarm notifications in time.

[0005] In existing vector autoregression cloud computing, it is impossible to analyze the stationarity and error data of related variables through virtual machine performance monitoring data to optimize the resource allocation and performance of the virtual machine, which increases the difficulty of detecting virtual machine performance anomalies. Summary of the Invention

[0006] In order to overcome the above-mentioned defects of the prior art, an embodiment of the present invention provides a method for monitoring virtual machines in cloud computing based on vector autoregression, which is used to solve the problems raised in the above-mentioned background technology.

[0007] To achieve the above objectives, the present invention provides the following technical solution: a method for monitoring virtual machines in cloud computing based on vector autoregression, comprising:

[0008] Step S01: Distributed monitoring node deployment: used to deploy monitoring nodes on the target virtual machine, define the data collection frequency of the monitoring nodes deployed on the target virtual machine, and number the data collection frequency of the monitoring nodes of the target virtual machine.

[0009] Step S02: virtual machine performance data collection: used to collect virtual machine performance data of each data collection frequency of the target virtual machine monitoring node, and the virtual machine performance data collection includes a virtual machine resource demand data collection unit and a virtual machine resource adjustment data collection unit.

[0010] Step S03: virtual machine resource demand status monitoring: used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, calculate the virtual machine resource demand overload index of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand data collection unit, and screen out high-load virtual machines.

[0011] Step S04: Virtual machine resource allocation policy management: used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, adjust the data collection unit according to the virtual machine resources to calculate the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, and perform high-load virtual machine resource migration.

[0012] Step S05: virtual machine performance optimization supervision: used to analyze and obtain the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand overload index and virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node.

[0013] Step S06: virtual machine performance optimization benefit evaluation: used to obtain the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource optimization supervision coefficient, and process it.

[0014] Step S07: Vector autoregression model virtual machine performance anomaly detection: Based on the variables virtual machine resource demand overload index, virtual machine load balancing distribution index and virtual machine resource optimization supervision coefficient, a vector autoregression model is used to verify each variable, obtain the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node, and verify the stability of virtual machine performance.

[0015] Preferably, the distributed monitoring node deployment is specifically as follows:

[0016] A monitoring node is deployed for the target virtual machine, and the monitoring nodes deployed for the target virtual machine are numbered 1, 2, ...j, ...m in sequence. The data collection frequency of the monitoring node deployed for the target virtual machine is defined, and the data collection frequency of the monitoring node of the target virtual machine is numbered 1, 2, ...i, ...n in sequence.

[0017] Preferably, the virtual machine performance data collection specifically includes:

[0018] Virtual machine resource demand data collection unit: collects the virtual machine response speed index and virtual machine data processing capacity index of each data collection frequency of the target virtual machine monitoring node, marked as Where i = 1, 2, ... n, i represents the number of the i-th data collection frequency, j = 1, 2, ... m, j represents the number of the j-th virtual machine monitoring node;

[0019] Virtual machine resource adjustment data collection unit: collects the virtual machine load evaluation and load balancer evaluation efficiency of each data collection frequency of the target virtual machine monitoring node, marked as

[0020] Preferably, the calculation formula of the virtual machine resource demand overload index is:

[0021]

[0022] in It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, It is expressed as the virtual machine response speed index of the jth virtual machine monitoring node i-th data collection frequency, Δqv j It is expressed as the mean value of the virtual machine response speed index of the j-th virtual machine monitoring node, and the formula is: It is expressed as the virtual machine data processing capability index of the jth virtual machine monitoring node with the i-th data collection frequency, Δqn j It is expressed as the mean value of the virtual machine data processing capability index of the j-th virtual machine monitoring node, and the formula is: μ1 and μ2 represent the correction factors of the virtual machine response speed index and the virtual machine data processing capability index, respectively, and n represents the total number of data collection frequencies;

[0023] Obtain the virtual machine resource demand overload index of the i-th data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource demand overload index, filter out the numbers of the virtual machine monitoring nodes whose virtual machine resource demand overload index is less than the preset virtual machine resource demand overload index, and mark them as high-load virtual machines.

[0024] Preferably, the virtual machine resource demand status monitoring further includes:

[0025] Step S001: Using a network packet capture tool to collect network delay, virtual machine request time, and virtual machine response time at each data collection frequency through a target virtual machine monitoring node;

[0026] Step S002: Calculate the virtual machine response speed index. The specific calculation formula is:

[0027]

[0028] in It is represented as the virtual machine response time of the jth virtual machine monitoring node with the i-th data collection frequency, It represents the virtual machine request time of the jth virtual machine monitoring node with the i-th data collection frequency, represents the preset network delay for the j-th virtual machine monitoring node, It is represented as the network delay of the jth virtual machine monitoring node at the i-th data collection frequency, and e is represented as a natural constant;

[0029] Step S003: Using a network packet capture tool to collect the disk read and write data volume, virtual machine error count, and packet loss rate at each data collection frequency through the target virtual machine monitoring node;

[0030] Step S004: Calculate the virtual machine data processing capability index. The specific calculation formula is:

[0031]

[0032] in It is expressed as the packet loss rate of the i-th data collection frequency of the j-th virtual machine monitoring node, It is expressed as the time of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented by the disk read and write data volume of the jth virtual machine monitoring node at the i-th data collection frequency, It represents the number of virtual machine errors of the i-th data collection frequency of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency.

[0033] Preferably, the calculation formula of the virtual machine load balancing distribution index is:

[0034]

[0035] in It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the VM load evaluation of the j-th VM monitoring node with the i-th data collection frequency, It represents the preset VM load evaluation of the j-th VM monitoring node, It is expressed as the load balancer evaluation efficiency of the j-th virtual machine monitoring node and the i-th data collection frequency, It is expressed as the minimum value of the load balancer evaluation efficiency of the i-th data collection frequency, It is represented as the maximum value of the load balancer evaluation efficiency at the i-th data collection frequency, and e is represented as a natural constant;

[0036] Obtain the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine load balancing distribution index, filter out the virtual machine load balancing distribution index greater than the preset virtual machine load balancing distribution index, mark it as a virtual machine that can receive resource allocation, and migrate the high-load virtual machine processing resources to the virtual machine that can receive resource allocation for processing.

[0037] Preferably, the virtual machine resource allocation policy management further includes:

[0038] Step S001: Using a network packet capture tool to monitor the target virtual machine node, collect the amount of data waiting for CPU processing and the number of I / O requests waiting for disk processing at each data collection frequency;

[0039] Step S002: Calculate the virtual machine load assessment. The specific calculation formula is:

[0040]

[0041] in It is represented as the maximum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the minimum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the amount of data waiting for CPU processing at the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the number of I / O requests waiting for disk processing at the i-th data collection frequency of the j-th virtual machine monitoring node;

[0042] Step S003: Using a network packet capture tool to collect the data transmission volume of the load balancer at each data collection frequency through the target virtual machine monitoring node;

[0043] Step S004: Calculate the load balancer evaluation efficiency. The specific calculation formula is:

[0044]

[0045] in It is represented as the data transmission volume of the load balancer of the jth virtual machine monitoring node with the i-th data collection frequency, It represents the data transmission volume of the preset load balancer of the j-th virtual machine monitoring node.

[0046] Preferably, the calculation formula of the virtual machine resource optimization supervision coefficient is:

[0047]

[0048] in It is expressed as the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It represents the maximum value of the virtual machine load balancing distribution index of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency.

[0049] Preferably, the virtual machine performance optimization benefit evaluation is specifically as follows:

[0050] Obtain the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node, and compare it with the preset virtual machine resource optimization supervision coefficient. If the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node is less than the preset virtual machine resource optimization supervision coefficient, it indicates that the resource optimization strategy of the virtual machine monitoring node does not meet expectations. In this case, the number of the virtual machine monitoring node is counted and an abnormal warning is issued. Otherwise, it indicates that the resource optimization strategy of the virtual machine monitoring node meets expectations.

[0051] Preferably, the vector autoregression model virtual machine performance anomaly detection is specifically as follows:

[0052] Get the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node. If the ω j <1, it indicates that the variables of the virtual machine monitoring node fluctuate significantly and the virtual machine performance stability is abnormal. Otherwise, it indicates that the variables of the virtual machine monitoring node have a stable trend and the virtual machine performance stability is normal.

[0053] Technical effects and advantages of the present invention:

[0054] 1. The present invention provides a method for monitoring virtual machines in cloud computing based on vector autoregression. The method collects virtual machine performance data of each data collection frequency of a target virtual machine monitoring node, calculates the virtual machine resource demand overload index of each data collection frequency of the target virtual machine monitoring node according to a virtual machine resource demand data collection unit, thereby screening out high-load virtual machines. The virtual machine resource adjustment data collection unit calculates the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, and performs high-load virtual machine resource migration. The virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node is further analyzed and compared with the preset virtual machine resource optimization supervision coefficient. If the virtual machine resource optimization supervision coefficient of the target virtual machine monitoring node at the i-th data collection frequency is less than the preset virtual machine resource optimization supervision coefficient, it indicates that the resource optimization strategy of the virtual machine monitoring node does not meet expectations, then the number of the virtual machine monitoring node is counted and an abnormal warning is issued. Otherwise, it indicates that the resource optimization strategy of the virtual machine monitoring node meets expectations. By collecting virtual machine performance data in real time and using this data to calculate the resource demand overload index and load balancing distribution index, future resource demand can be predicted more accurately, which helps to reasonably allocate tasks on high-load virtual machines to other low-load virtual machines, thereby maintaining the stable operation of the entire system and adopting monitoring strategies to respond to resource allocation abnormalities in a timely manner;

[0055] 2. The present invention provides a method for monitoring virtual machines in cloud computing based on vector autoregression. The method detects virtual machine performance anomalies based on a vector autoregression model. Based on the variables virtual machine resource demand overload index, virtual machine load balancing distribution index and virtual machine resource optimization supervision coefficient, the vector autoregression model is used to verify each variable to obtain the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node. If the ω of a certain virtual machine monitoring node j <1, it indicates that the variables of the virtual machine monitoring node fluctuate significantly and there are abnormalities in the virtual machine performance stability. Conversely, it indicates that the trends of the variables of the virtual machine monitoring node are stable and there are no abnormalities in the virtual machine performance stability. The vector autoregression model is a time series analysis model that can capture the dynamic correlation between variables. In virtual machine performance testing, this dynamic correlation analysis helps to understand the mutual influence between different performance indicators, thereby more accurately predicting and identifying performance stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 It is a schematic diagram of the connection of the method steps of the present invention.

[0057] Figure 2 This is a schematic diagram of the virtual machine performance data collection structure of the present invention. DETAILED DESCRIPTION

[0058] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0059] See also Figure 1 As shown, the present invention provides a virtual machine monitoring method in cloud computing based on vector autoregression, which includes the following steps: distributed monitoring node deployment, virtual machine performance data collection, virtual machine resource demand status monitoring, virtual machine resource allocation strategy management, virtual machine performance optimization supervision, virtual machine performance optimization benefit evaluation, and vector autoregression model virtual machine performance anomaly detection.

[0060] The distributed monitoring node deployment is connected with virtual machine performance data collection, virtual machine performance data collection is connected with virtual machine resource demand status monitoring, virtual machine resource demand status monitoring is connected with virtual machine resource allocation policy management, virtual machine resource allocation policy management is connected with virtual machine performance optimization supervision, virtual machine performance optimization supervision is connected with virtual machine performance optimization benefit evaluation, and virtual machine performance optimization benefit evaluation is connected with vector autoregression model virtual machine performance anomaly detection.

[0061] The step S01: distributed monitoring node deployment: is used to deploy monitoring nodes on the target virtual machine, define the data collection frequency of the monitoring nodes deployed on the target virtual machine, and number the data collection frequency of the monitoring nodes of the target virtual machine.

[0062] In one possible design, the distributed monitoring nodes are deployed as follows:

[0063] A monitoring node is deployed for the target virtual machine, and the monitoring nodes deployed for the target virtual machine are numbered 1, 2, ...j, ...m in sequence. The data collection frequency of the monitoring node deployed for the target virtual machine is defined, and the data collection frequency of the monitoring node of the target virtual machine is numbered 1, 2, ...i, ...n in sequence.

[0064] See also Figure 2 As shown, the step S02: virtual machine performance data collection: is used to collect virtual machine performance data of each data collection frequency of the target virtual machine monitoring node, and the virtual machine performance data collection includes a virtual machine resource demand data collection unit and a virtual machine resource adjustment data collection unit.

[0065] In one possible design, the virtual machine performance data collection is specifically as follows:

[0066] Virtual machine resource demand data collection unit: collects the virtual machine response speed index and virtual machine data processing capacity index of each data collection frequency of the target virtual machine monitoring node, marked as Where i = 1, 2, ... n, i represents the number of the i-th data collection frequency, j = 1, 2, ... m, j represents the number of the j-th virtual machine monitoring node;

[0067] Virtual machine resource adjustment data collection unit: collects the virtual machine load evaluation and load balancer evaluation efficiency of each data collection frequency of the target virtual machine monitoring node, marked as

[0068] The step S03: virtual machine resource demand status monitoring: is used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, calculate the virtual machine resource demand overload index of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand data collection unit, and screen out high-load virtual machines.

[0069] In one possible design, the virtual machine resource demand status monitoring is specifically as follows:

[0070] Step S001: Using a network packet capture tool to collect network delay, virtual machine request time, and virtual machine response time at each data collection frequency through a target virtual machine monitoring node;

[0071] Step S002: Calculate the virtual machine response speed index. The specific calculation formula is:

[0072]

[0073] in It is represented as the virtual machine response speed index of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the virtual machine response time of the jth virtual machine monitoring node with the i-th data collection frequency, It represents the virtual machine request time of the jth virtual machine monitoring node with the i-th data collection frequency, represents the preset network delay for the j-th virtual machine monitoring node, It is represented as the network delay of the jth virtual machine monitoring node at the i-th data collection frequency, and e is represented as a natural constant;

[0074] Step S003: Using a network packet capture tool to collect the disk read and write data volume, virtual machine error count, and packet loss rate at each data collection frequency through the target virtual machine monitoring node;

[0075] Step S004: Calculate the virtual machine data processing capability index. The specific calculation formula is:

[0076]

[0077] in It is expressed as the virtual machine data processing capability index of the i-th data collection frequency of the j-th virtual machine monitoring node, It is expressed as the packet loss rate of the i-th data collection frequency of the j-th virtual machine monitoring node, It is expressed as the time of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented by the disk read and write data volume of the jth virtual machine monitoring node at the i-th data collection frequency, It represents the number of virtual machine errors of the i-th data collection frequency of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency;

[0078] Step S005: The calculation formula of the virtual machine resource demand overload index is:

[0079]

[0080] in It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, Δqv j It is expressed as the mean value of the virtual machine response speed index of the j-th virtual machine monitoring node, and the formula is: Δqn j It is expressed as the mean value of the virtual machine data processing capability index of the j-th virtual machine monitoring node, and the formula is: μ1 and μ2 represent the correction factors of the virtual machine response speed index and the virtual machine data processing capability index, respectively, and n represents the total number of data collection frequencies;

[0081] Step S006: Obtain the virtual machine resource demand overload index of the i-th data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource demand overload index, filter out the numbers of the virtual machine monitoring nodes whose virtual machine resource demand overload index is less than the preset virtual machine resource demand overload index, and mark them as high-load virtual machines.

[0082] The step S04: virtual machine resource allocation policy management: is used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, adjust the data collection unit according to the virtual machine resources to calculate the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, and perform high-load virtual machine resource migration.

[0083] In one possible design, the virtual machine resource allocation policy management is specifically as follows:

[0084] Step S001: Using a network packet capture tool to monitor the target virtual machine node, collect the amount of data waiting for CPU processing and the number of I / O requests waiting for disk processing at each data collection frequency;

[0085] Step S002: Calculate the virtual machine load assessment. The specific calculation formula is:

[0086]

[0087] in It is represented as the VM load evaluation of the j-th VM monitoring node with the i-th data collection frequency, It is represented as the maximum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the minimum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the amount of data waiting for CPU processing at the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the number of I / O requests waiting for disk processing at the i-th data collection frequency of the j-th virtual machine monitoring node;

[0088] Step S003: Using a network packet capture tool to collect the data transmission volume of the load balancer at each data collection frequency through the target virtual machine monitoring node;

[0089] Step S004: Calculate the load balancer evaluation efficiency. The specific calculation formula is:

[0090]

[0091] in It is expressed as the load balancer evaluation efficiency of the j-th virtual machine monitoring node and the i-th data collection frequency, It is represented as the data transmission volume of the load balancer of the jth virtual machine monitoring node with the i-th data collection frequency, represents the data transmission volume of the preset load balancer of the j-th virtual machine monitoring node;

[0092] Step S005: The calculation formula of the virtual machine load balancing distribution index is:

[0093]

[0094] in It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It represents the preset VM load evaluation of the j-th VM monitoring node, It is expressed as the minimum value of the load balancer evaluation efficiency of the i-th data collection frequency, It is represented as the maximum value of the load balancer evaluation efficiency at the i-th data collection frequency, and e is represented as a natural constant;

[0095] Step S006: Obtain the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine load balancing distribution index, filter out the virtual machine load balancing distribution index greater than the preset virtual machine load balancing distribution index, mark it as a virtual machine that can receive resource allocation, and migrate the high-load virtual machine processing resources to the virtual machine that can receive resource allocation for processing.

[0096] The step S05: virtual machine performance optimization supervision: is used to analyze and obtain the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand overload index and virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node.

[0097] In one possible design, the calculation formula for the virtual machine resource optimization supervision coefficient is:

[0098]

[0099] in It is expressed as the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It represents the maximum value of the virtual machine load balancing distribution index of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency.

[0100] The step S06: virtual machine performance optimization benefit evaluation: is used to obtain the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource optimization supervision coefficient, and process it.

[0101] In one possible design, the virtual machine performance optimization benefit evaluation is specifically as follows:

[0102] Obtain the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node, and compare it with the preset virtual machine resource optimization supervision coefficient. If the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node is less than the preset virtual machine resource optimization supervision coefficient, it indicates that the resource optimization strategy of the virtual machine monitoring node does not meet expectations. In this case, the number of the virtual machine monitoring node is counted and an abnormal warning is issued. Otherwise, it indicates that the resource optimization strategy of the virtual machine monitoring node meets expectations.

[0103] Step S07: Vector autoregression model virtual machine performance anomaly detection: Based on the variables virtual machine resource demand overload index, virtual machine load balancing distribution index and virtual machine resource optimization supervision coefficient, a vector autoregression model is used to verify each variable, and the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node is obtained to verify the stability of virtual machine performance.

[0104] In a possible design, the vector autoregression model virtual machine performance anomaly detection is specifically as follows:

[0105] Step S001: Calculate the virtual machine performance anomaly detection coefficient:

[0106]

[0107] where ω j is the virtual machine performance anomaly detection coefficient of the jth virtual machine monitoring node, γ is a constant vector, m is the number of virtual machine monitoring nodes, σ α Expressed as the standard deviation of the virtual machine resource demand overload index, σ β It is represented as the standard deviation of the virtual machine load balancing distribution index, and p is the lag order, which is 2.

[0108] Step S002: Obtain the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node. If the ω j <1, it indicates that the variables of the virtual machine monitoring node fluctuate significantly and the virtual machine performance stability is abnormal. Otherwise, it indicates that the variables of the virtual machine monitoring node have a stable trend and the virtual machine performance stability is normal.

[0109] In this embodiment, it should be specifically explained that the present invention collects the virtual machine performance data of each data collection frequency of the target virtual machine monitoring node, calculates the virtual machine resource demand overload index of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand data collection unit, thereby screening out high-load virtual machines, and adjusts the virtual machine resource collection unit to calculate the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, and performs high-load virtual machine resource migration, further analyzes and obtains the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node, and compares it with the preset virtual machine resource optimization supervision coefficient. If the target virtual machine If the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the virtual machine monitoring node is less than the preset virtual machine resource optimization supervision coefficient, it indicates that the resource optimization strategy of the virtual machine monitoring node does not meet expectations, then the number of the virtual machine monitoring node is counted and an abnormal warning is issued. Otherwise, it indicates that the resource optimization strategy of the virtual machine monitoring node meets expectations. By collecting virtual machine performance data in real time and using this data to calculate the resource demand overload index and load balancing distribution index, future resource requirements can be predicted more accurately, which helps to reasonably allocate tasks on high-load virtual machines to other low-load virtual machines, thereby maintaining the stable operation of the entire system and adopting monitoring strategies to respond to resource allocation anomalies in a timely manner.

[0110] In this embodiment, it should be specifically explained that the present invention is based on the vector autoregression model virtual machine performance anomaly detection, based on the variables virtual machine resource demand overload index, virtual machine load balancing distribution index and virtual machine resource optimization supervision coefficient, using the vector autoregression model to verify each variable, and obtain the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node. If the ω of a certain virtual machine monitoring node j <1, it indicates that the variables of the virtual machine monitoring node fluctuate significantly and there are abnormalities in the virtual machine performance stability. Conversely, it indicates that the trends of the variables of the virtual machine monitoring node are stable and there are no abnormalities in the virtual machine performance stability. The vector autoregression model is a time series analysis model that can capture the dynamic correlation between variables. In virtual machine performance testing, this dynamic correlation analysis helps to understand the mutual influence between different performance indicators, thereby more accurately predicting and identifying performance stability.

[0111] Finally: The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for monitoring virtual machines in cloud computing based on vector autoregression, characterized in that: include: Step S01: Distributed monitoring node deployment: used to deploy monitoring nodes on the target virtual machine, define the data collection frequency of the monitoring nodes deployed on the target virtual machine, and number the data collection frequency of the monitoring nodes of the target virtual machine; Step S02: Virtual machine performance data collection: used to collect virtual machine performance data of each data collection frequency of the target virtual machine monitoring node, the virtual machine performance data collection includes a virtual machine resource demand data collection unit and a virtual machine resource adjustment data collection unit; Step S03: Virtual machine resource demand status monitoring: used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, calculate the virtual machine resource demand overload index of each data collection frequency of the target virtual machine monitoring node according to the virtual machine resource demand data collection unit, and screen out high-load virtual machines; Step S04: Virtual machine resource allocation policy management: used to receive the virtual machine performance data transmitted by the virtual machine performance data collection step, adjust the data collection unit according to the virtual machine resources to calculate the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, and perform high-load virtual machine resource migration; Step S05: virtual machine performance optimization supervision: for analyzing and obtaining a virtual machine resource optimization supervision coefficient for each data collection frequency of the target virtual machine monitoring node based on the virtual machine resource demand overload index and the virtual machine load balancing distribution index for each data collection frequency of the target virtual machine monitoring node; Step S06: virtual machine performance optimization benefit evaluation: used to obtain the virtual machine resource optimization supervision coefficient of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource optimization supervision coefficient, and process it; Step S07: Vector autoregression model virtual machine performance anomaly detection: Based on the variables virtual machine resource demand overload index, virtual machine load balancing distribution index and virtual machine resource optimization supervision coefficient, a vector autoregression model is used to verify each variable, obtain the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node, and verify the stability of virtual machine performance.

2. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 1, characterized in that: The distributed monitoring node deployment is specifically as follows: A monitoring node is deployed for the target virtual machine, and the monitoring nodes deployed for the target virtual machine are numbered 1, 2, ...j, ...m in sequence. The data collection frequency of the monitoring node deployed for the target virtual machine is defined, and the data collection frequency of the monitoring node of the target virtual machine is numbered 1, 2, ...i, ...n in sequence.

3. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 2, characterized in that: The virtual machine performance data collection is specifically as follows: Virtual machine resource demand data collection unit: collects the virtual machine response speed index and virtual machine data processing capacity index of each data collection frequency of the target virtual machine monitoring node, marked as Where i = 1, 2, ... n, i represents the number of the i-th data collection frequency, j = 1, 2, ... m, j represents the number of the j-th virtual machine monitoring node; Virtual machine resource adjustment data collection unit: collects the virtual machine load evaluation and load balancer evaluation efficiency of each data collection frequency of the target virtual machine monitoring node, marked as 4. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 3, characterized in that: The calculation formula of the virtual machine resource demand overload index is: in It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, It is expressed as the virtual machine response speed index of the jth virtual machine monitoring node i-th data collection frequency, Δqv j It is represented as the mean value of the virtual machine response speed index of the j-th virtual machine monitoring node, It is expressed as the virtual machine data processing capability index of the jth virtual machine monitoring node with the i-th data collection frequency, Δqn j It is represented as the mean value of the virtual machine data processing capability index of the jth virtual machine monitoring node, μ1 and μ2 are the correction factors of the virtual machine response speed index and the virtual machine data processing capability index respectively, and n is the total number of data collection frequencies; Obtain the virtual machine resource demand overload index of the i-th data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine resource demand overload index, filter out the numbers of the virtual machine monitoring nodes whose virtual machine resource demand overload index is less than the preset virtual machine resource demand overload index, and mark them as high-load virtual machines.

5. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 4, characterized in that: The virtual machine resource demand status monitoring also includes: Step S001: Using a network packet capture tool to collect network delay, virtual machine request time, and virtual machine response time at each data collection frequency through a target virtual machine monitoring node; Step S002: Calculate the virtual machine response speed index. The specific calculation formula is: in It is represented as the virtual machine response time of the jth virtual machine monitoring node with the i-th data collection frequency, It represents the virtual machine request time of the jth virtual machine monitoring node with the i-th data collection frequency, represents the preset network delay for the j-th virtual machine monitoring node, It is represented as the network delay of the jth virtual machine monitoring node at the i-th data collection frequency, and e is represented as a natural constant; Step S003: Using a network packet capture tool to collect the disk read and write data volume, virtual machine error count, and packet loss rate at each data collection frequency through the target virtual machine monitoring node; Step S004: Calculate the virtual machine data processing capability index. The specific calculation formula is: in It is expressed as the packet loss rate of the i-th data collection frequency of the j-th virtual machine monitoring node, It is expressed as the time of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented by the disk read and write data volume of the jth virtual machine monitoring node at the i-th data collection frequency, It represents the number of virtual machine errors of the i-th data collection frequency of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency.

6. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 5, characterized in that: The calculation formula of the virtual machine load balancing distribution index is: in It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the VM load evaluation of the j-th VM monitoring node with the i-th data collection frequency, It represents the preset VM load evaluation of the j-th VM monitoring node, It is expressed as the load balancer evaluation efficiency of the j-th virtual machine monitoring node and the i-th data collection frequency, It is expressed as the minimum value of the load balancer evaluation efficiency of the i-th data collection frequency, It is represented as the maximum value of the load balancer evaluation efficiency at the i-th data collection frequency, and e is represented as a natural constant; Obtain the virtual machine load balancing distribution index of each data collection frequency of the target virtual machine monitoring node, compare it with the preset virtual machine load balancing distribution index, filter out the virtual machine load balancing distribution index greater than the preset virtual machine load balancing distribution index, mark it as a virtual machine that can receive resource allocation, and migrate the high-load virtual machine processing resources to the virtual machine that can receive resource allocation for processing.

7. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 6, characterized in that: The virtual machine resource allocation policy management also includes: Step S001: Using a network packet capture tool to monitor the target virtual machine node, collect the amount of data waiting for CPU processing and the number of I / O requests waiting for disk processing at each data collection frequency; Step S002: Calculate the virtual machine load assessment. The specific calculation formula is: in It is represented as the maximum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the minimum value of the virtual machine response speed index of the i-th data collection frequency, It is represented as the amount of data waiting for CPU processing at the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the number of I / O requests waiting for disk processing at the i-th data collection frequency of the j-th virtual machine monitoring node; Step S003: Using a network packet capture tool to collect the data transmission volume of the load balancer at each data collection frequency through the target virtual machine monitoring node; Step S004: Calculate the load balancer evaluation efficiency. The specific calculation formula is: in It is represented as the data transmission volume of the load balancer of the jth virtual machine monitoring node with the i-th data collection frequency, It represents the data transmission volume of the preset load balancer of the j-th virtual machine monitoring node.

8. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 7, characterized in that: The calculation formula of the virtual machine resource optimization supervision coefficient is: in It is expressed as the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the j-th virtual machine monitoring node, It is represented as the virtual machine resource demand overload index of the jth virtual machine monitoring node with the i-th data collection frequency, It is represented as the virtual machine load balancing distribution index of the i-th data collection frequency of the j-th virtual machine monitoring node, It represents the maximum value of the virtual machine load balancing distribution index of the j-th virtual machine monitoring node, and i-1 represents the number of the i-1-th data collection frequency.

9. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 8, characterized in that: The virtual machine performance optimization benefit evaluation is specifically as follows: Obtain the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node, and compare it with the preset virtual machine resource optimization supervision coefficient. If the virtual machine resource optimization supervision coefficient of the i-th data collection frequency of the target virtual machine monitoring node is less than the preset virtual machine resource optimization supervision coefficient, it indicates that the resource optimization strategy of the virtual machine monitoring node does not meet expectations. In this case, the number of the virtual machine monitoring node is counted and an abnormal warning is issued. Otherwise, it indicates that the resource optimization strategy of the virtual machine monitoring node meets expectations.

10. The method for monitoring virtual machines in cloud computing based on vector autoregression according to claim 9, characterized in that: The vector autoregression model virtual machine performance anomaly detection is specifically as follows: Get the virtual machine performance anomaly detection coefficient of the target virtual machine monitoring node. If the ω j <1, it indicates that the variables of the virtual machine monitoring node fluctuate significantly and the virtual machine performance stability is abnormal. Otherwise, it indicates that the variables of the virtual machine monitoring node have a stable trend and the virtual machine performance stability is normal.

Images