Method for determining virtual machine migration opportunity based on prediction of utilization rate change trend

Abstract

The invention provides a new method for determining a virtual machine migration opportunity. The method comprises the following steps: when any resource utilization rate of a virtual machine reaches acertain value, starting a prediction module; performing numerical fitting according to the resource utilization rates corresponding to an interval fixed time length within the closer time before themoment; deducing the function to obtain a function corresponding to a change slope; solving a change value of the function in a future period of time to obtain a change trend of the utilization rate;if the change trend is always increasing, determining as the virtual machine migration opportunity; if the change trend is in a decreasing state, continuously monitoring the utilization rate; if the change trend is in a fluctuation state, calculating an integral value of a slope function in the future period of time to obtain the change value of the utilization rate; and comparing the change valuewith a set threshold value to determine the migration opportunity. After the prediction stage, if the utilization rate is still increasing and reaches a limit value, the duration time of the utilization rate reaching the limit value is counted as a judgment condition of virtual machine migration. According to the method, by predicting the change trend of each utilization rate of the virtual machine, a tight resource is located; corresponding judgments are made according to different change trends to implement accurate positioning of the migration opportunity.

Description

(1) Technical field [0001] The invention relates to the field of cloud computing, in particular to a method for migrating a virtual machine based on prediction of utilization rate variation trend in a cloud environment. (2) Background technology [0002] With the increasing request for computer resources at different levels, cloud computing emerged as a large-scale distributed computing model. With the help of virtualization technology, cloud computing realizes resource management at the granularity of virtual machines, and virtual machines execute tasks submitted by users by sharing resources on physical machines. However, the changing task requests submitted by users may easily cause resource competition among virtual machines, which cannot guarantee server-level agreements and user service quality. At this time, by migrating high-load physical virtual machines to light-load physical machines, resource Redeploy to effectively avoid resource competition. [0003] The key ...

AI Technical Summary

Problems solved by technology

[0003] The key technology of virtual machine migration includes the determination of virtual machine migration timing. Inaccurate virtual machine migration timing will lead to invalid migration or frequent migration, which will generate a lot of energy consumption

The current virtual machine migration is mainly in the passive stage, and the timing of migration is after resource competition occurs, there is a certain lag, and the user service quality cannot be truly guaranteed

In the latest research, a method of predicting the number of task requests for active migration is proposed to solve this problem, but it ignores the different requirements of tasks for different resources, and it is impossible to locate tight resources. Migration of virtual machines based on this may easily lead to invalid migration

Images