Optimizing the Cost for Resource Subscription Policy in IaaS Cloud (1402.2491v1)

Abstract: Cloud computing allow the users to efficiently and dynamically provision computing resource to meet their IT needs. Cloud Provider offers two subscription plan to the customer namely reservation and on-demand. The reservation plan is typically cheaper than on-demand plan. If the actual computing demand is known in advance reserving the resource would be straightforward. The challenge is how to make properly resource provisioning and how the customers efficiently purchase the provisioning options under reservation and on-demand. To address this issue, two-phase algorithm are proposed to minimize service provision cost in both reservation and on-demand plan. To reserve the correct and optimal amount of resources during reservation, proposed a mathematical formulae in the first phase. To predict resource demand, use kalman filter in the second phase. The evaluation result shows that the two-phase algorithm can significantly reduce the provision cost and the prediction is of reasonable accuracy.