The Research Of Resource Provisioning Based On Hungarian Algorithm In Cloud Computing

In order to meet the growing demand for computing, storage and other resources,the size of the data center increasing bigger and more complex, the development of theIT industry is also beginning to change from mode of the saling hardware and softwareresources to cloud computing model of providing users with the infrastructure, platform,and software services, so how cloud computing to provide users with reliable servicehas become the focus of the industry. In the process of providing cloud services,resource provisioning is an important and fundamental aspect. Resource provisioning incloud computing refers to reasonably divide virtual machines on the physical entity forusers. Domestic and foreign research about resource provisioning, however, is notsufficient, and some of the major cloud service providers of its resource provisioningstrategy implementation is not public, and most of the literature is based on singleoptimization objective. Therefore, to solve the above problems, resource provisioningstrategy of cloud computing will be carried out in-depth in the following research.Major work:The resource provisioning strategy based on Hungarian algorithm has beenproposed, the strategy has a main objective: Make sure the execution time of the groupof user’s jobs is the least; as the same time ensure that each user is assigned resources tomeet user tasks’ QoS requirements, and keep load balancing, make the tasks evenlydistributed to the physical host in the resource pool.This thesis has proposed two rescheduling algorithm of resources, the target of thefirst algorithm is as little as possible to migrate virtual machine, the other algorithm is tominimize the overhead of the migration, first select the relatively low overhead one ofthe virtual machine to migrate. Though the objectives of the two algorithms are different,they both can reduce the energy consumption of the entire system.Use Matlab mathematical software to realize the resource provisioning strategybased on Hungarian algorithm and simulate through CloudSim platform. Then in allthree cases: The number of user requests is equal to the number of resources, the number of user requests is greater than the number of resources, and the number of userrequests is less than the number of resources and users; has made resource provisioningstrategy based on Hungarian algorithm compare with resource provisioning based onMin-Min algorithm, the result has shown that the resource provisioning strategy basedon Hungarian algorithm proposed in this thesis make the tasks’ execution time less thanthe resource provisioning based on Min-Min algorithm, then verify the correctness andvalidity of the proposed algorithm. Finally, comparing the two re-scheduling algorithmboth of their energy saving and the number of virtual machine migration, experimentshas proved that the algorithm of as little as possible to migrate virtual machine not onlysave more energy consumption, but also the less migration.