Deadline based Virtual Machine Provisioning to improve QoS in Cloud Computing

Cloud computing delivers IT resources as a Service over Internet. It is growing as the future for application deployment depending upon Quality of Service provided by the service provider. Virtualization provides the base for the same; it has certain pros like portability, efficient hardware utilization, ease in maintenance, cost benefit and cons like performance degradation if virtualized resources are not uniformly allocated. In this paper user requests are bifurcated on the bases of type of deadline (unlimited or tight). To meet Tight Deadline requirements we have created Heterogeneous Cloud Computing Environment (virtual machines with different configuration are dynamically created and destroyed). CloudSim toolkit is used to simulate the results of Heterogeneous Cloud Environment over a node based on deadline of incoming user request. We show that our approach improves request rejection ratio, deadline miss rate, response time to user and resource utilization. Ultimately Quality of Service (QoS) is improved.