A service-oriented architecture for on-demand and dynamic ally reconfigurable computing

As the performance and size of computing facilities and associated data services continue to increase, the management of computing systems within data centers is becoming a significant challenge. This paper analyzes the characteristics of services in the Scientific Data Center of the Chinese Academy of Sciences. To capture the service characteristics, the paper describes three computing system architectures and automatic deployment methods for three types of services. Specifically, aiming at facilitating management, guaranteeing QoS, improving resource utilization and relieving energy consumption, the paper proposes a novel computing system architecture for SDC, on- demand service-oriented dynamic reconfigurable computing system (SODRCS). Based on the proposed architectures, automatic deployment methods and the policies of assigning/reclaiming resources, SODRCS enables on-demand resource provision by automatically reconfiguring computing systems with the variety of service demand.