Hybrid Resource Provisioning Method For Workflow Scheduling In Cloud Computing

In cloud computing,Cloud service providers always provide two provisioning plans to cloud consumers for computing resources,namely reservation and on-demand plans.In general,cost of the resources provisioned by reservation plan is cheaper than on-demand plan. however, the re-sources provisioned by reservation must be utilized throughout the computing period,so maybe it’s a waste.The cost maybe reduced by renting resources from both that two plans.Therefore, Hybrid resource provisioning for workflow scheduling in cloud computing which is full of the-oretic significance and practical value is considered in the thesis.Considered the basic features of workflow in cloud computing, a mathematics model is established. To address this problem, a heuristic algorithm is proposed, which is consisted of three phases:generating initial solution, optimizing schedule sequence, solution rebuilding. The algorithm of generating initial solution first generate a schedule sequence using an existing method, and allocate tasks on the reservation VMs in turn, checking if on-demand resource can help speed the tasks and reduce the cost. A local search algorithm is proposed to adjust the sequence of task scheduling, optimizing the initial solution. And a solution rebuilding strategy is proposed to improve the diversity of solution and optimize the solution, by gradually reducing the amount of on-demand resource.Experiment results show that MEFT algorithm can get better solutions than ABTS. In dif-ferent scales of instances and different discounts of reservation situation, MEFT can save more cost. And the huger the instance scales, the more cost can be saved.