Research And Implementation Of Resource Scheduling Algorithm For Virtual Machine In Cloud Computing

Virtualization technology has the advantage of enhancing system flexibility and scalability, improving resource utilization and meeting flexible application requirements that becomes an important technology. And virtual machines resource scheduling technology is one of the key technologies in the field. How to design virtual machines resource scheduling algorithm that ensures load balancing in a cloud platform and improves user experience, it already becomes a current research hot. Currently most of virtual machines resource scheduling algorithms process requests submitted by users serializely, and that may lead to the overall user bad experience and load imbalance in the cloud platform. Therefore, this thesis relies on the National Marine Cloud Computing project, and deeply research virtual machines resource scheduling algorithms. The main contributions can be summarized as follows:(1) Currently most of resources scheduling algorithms for virtual machines have two problems that the overall completion time is long for the deployment of batch jobs and load imbalance in a cloud platform, this thesis proposes virtual machines resource scheduling algorithms based on PC-BC (Protential capacity and Balance capacity). This algorithm comprehensively considers the overall job completion time and cloud platform load balance, and it uses the two metrics of potential capacity and balance capacity to measure virtual machines remaining capacity. On this basis this thesis designs batch jobs match scheduling algorithms, this algorithm caculates PC-BC matrix, resource matrix and weight matrix to get value matrix; Finally uses cross destrutive method and value evaluation standard to deploy batch jobs;(2) Moreover some resource scheduling algorithms for virtual machines have virtual machines instability and time not cost-effective problems, this thesis proposes virtual machines resource scheduling algorithms base on QoS constraint. This algorithm comphrensively considers virtual machines credibility and premium services, and it applys time memership function, cost memership function, credibility evaluation model and two weight vectors to cacualte comprehensive benefit value, and finally deploy jobs according to the comprehensive benefit value;(3) Finally, this thesis virifiys performance through experiment virtual machines resource scheduling algorithms based on PC-BC matrixs and QoS contriants. The experimental results show that PC-BC algorithm experiment has achieved good results on the overall job completion time and load balancing in a cloud platform; QoS constriants experiment has also achieved good results on the numbers of jobs completion and comprehensive benefits value.