Personal Virtualization Strategy Oriented Task Classify In Cloud Computing

How to improve the resource utilization is one of the points in cloud computingresearches. There is a way to improve the resource utilization through predicting theresource requirements of the tasks needed by artificial intelligence and choosing abest plan of tasks allocation by the servers’ resource utilization. However, in thecloud environment that with a huge number of tasks and take virtual machines asresource scheduling unit, this way is not only platform specified, but also will costmore energy to collect, store and compute data, integrate and migrate virtualmachines.In this paper, we have proposed a strategy that oriented task classify andpersonal virtualization in the cloud computing environment. Through comprehensiveanalyzing the tasks and integrating the communication protocol, programmeasurement, resource prediction and some analyze way to estimate the requirementfeatures of processor, network, disk resource that the task needed. And then,classifying the tasks to processor bound task, communicate bound task, disk boundtask, processor-communicate bound task, processor-disk bound task,communicate-disk bound task, strength standard task and weaker standard task bythe resource features of the tasks cost. After classified tasks, we can virtualizenumbers of personalized virtual machines that corresponding the task types byassign specific processor, network and disk resources with the virtualizationtechnology. Then, we proposed a personal virtualization strategy that aim atclassified task and personalized virtual machines. The strategy, with certainconstraints, will create new virtual machines or recycle the old virtual machines andstart up or shut down the physical machine when needed. Compare to the traditionalway that base on resource prediction and load balance, this strategy can avoid theresource unbalanced situation that some resources shortage and some superfluous.And also can reduce the times of virtual machines integration or migration at thesame time. Finally, through experiments verifying the validity of the task classify system insimulate cloud computing environment with different hardware configuration andcompare the resource utilization with the personal virtualization strategy and withoutit. The result shows that our strategy can improve the resource utilization efficientlyin the cloud computing environment.