Resource Scheduling In Hybrid Cloud

Workflow technology has been widely applied in assembly line production, office automation, scientific research, and other fields. With the popularity of cloud computing, the scheduling of workflow tasks in a cloud environment has become a new topic of workflow management systems. As the cloud computing is based on utility and consumption of computer resources, the research of workflow scheduling has shifted from time-based strategy to QoS-based strategy. Hybrid clouds, as a new cloud environment, has great practical value. Hybrid cloud resource allocation has become a research hot spot in recent years.The existing workflow scheduling algorithms are based on gird or single cloud. Scheduling workflow on hybrid cloud need to consider more factors, such as task assignment, resource selection, etc. What’s more, the existing hybrid cloud scheduling algorithms are time intensive. In this paper, we analyze the situation of workflow scheduling in hybrid cloud, present a scheduling model with data sensibility, deadline and cost constraints. Then, in order to minimize the monetary cost, we use integer linear program(ILP) to formulate the scheduling problem. We also propose the task assignment filter strategy based on Pareto Optimality to decrease the running time.The main research work and contribution of this paper is as follows:1. We study the current research result of resource scheduling in cloud, and focus on the workflow scheduling based on performance and cost scheduling strategy. By analyzing the structure and business model of hybrid cloud, and combine with two strategy of workflow scheduling, best-effort and QoS, we give out the hybrid cloud workflow scheduling model under the data sensibility, deadline and cost constraints.2. We consider two situations of the scheduling problem, the private cloud only or the hybrid cloud. The PCH(Path Cluster Heuristic) approach is used to judge whether the private cloud can meet the user’s requirements. When private cloud power is enough, the PCP(Partial Critical Path) algorithm is applied. ILP is used to formulate the problem in hybrid cloud.3. In order to improve the running time of ILP, the task assignment configurations should be optimized. By using the Pareto Optimality theory to remove the task assignment configuration which cannot be the optimal solution before resource allocation, we propose a task assignment filter strategy. In this way, the mapping relationship between tasks and computing resources is decreased. In the procedure of ILP, the time and cost constraints are considered.4. Simulation and Analysis. Our simulation shows that task assignment optimization can decrease the scheduling running time with a good result.