Research of SLA-Based Multitask-User-RequestsAdmission Control and Related Algorithm for theCloud Service Provider

摘要

With the population of Cloud service users showing an explosive growth, how to reduce costs and improve resource utilization while ensuring service quality has become a very important bargaining chip for the cloud service provider can survive in the fierce competition in the industry or not. Based on the service level agreement (SLA) constraints of Cloud service, this paper studies the request admission control strategy of multitask-user-requests. By discussing and analyzing the possible problems in the different stages of request admission, resource scheduling and subtasks execution for multitask-user-requests, tow priority dynamic configuration strategies and a delay compensation strategy are proposed to improve the ratio of requests acceptance, cut down SLA default rates during subtasks are executed, and improve users’ experience while the SLA violate occurs. What’s more, two resource scheduling strategies aimed at improving the resource utilization rate to reduce the cost of Cloud services as much as possible are proposed. And at the end of the article put forward the Dynamic configuring Priority Resources Scheduling (DPS) algorithm which based on the strategies above, and designed the relevant comparative experiments to verify the algorithm of DPS. The results showed that the DPS algorithm can reduce the usage of resources by improving resource utilization rate and helps the service providers to create much more profits than the other two algorithms to a certain extent.