The Research Of Grid Resource Scheduling Based On Multi-QoS Requirements

Grid is a kind of technology and management method that can combine the billions of computing resources, storage resources, data resources, information resources and knowledge resources on the network to form a virtual logical unit, and transform the resources into some an available, reliable, economical computing power. Resource scheduling is especially important in the research of grid technology. It can ensure the grid users’tasks to be allocated to the corresponding grid resources according to the computing capability and resource status of grid nodes. At the same time, the grid users’QoS (Quality of Service) requirements are satisfied.The resource scheduling problem is a NP complete problem, studies show that such problem is often difficult to gain the optimal solution, or even simply it doesn’t exist. The paper introduced the satisfaction function to search the satisfactory solution under the grid users’multi-QoS requirements, and classified the numerous QoS (Quality of Service) parameters into two types:performance QoS and trust QoS. the former included priority, timeliness and precision, and the latter included security and reliability. At the same time, the service level was divided into three:hard, soft and try. Firstly, the paper characterized and quantified each dimension of QoS constraint indexes; Then it constructed five QoS satisfaction function models and formulated the overall satisfaction function model; At last, based on the grid users’satisfaction requirements and the grid resources’satisfaction guarantee, the paper put forward the improved Q-Min-Min algorithm combined with Min-Min algorithm, and replaced the ETC (Expected Time to Compute) matrix in the classic Min-Min algorithm with the QSM (QoS Satisfied Matrix)to solve the sole consideration of makespan. So the problem of resource scheduling based on multi-QoS requirements was researched.The simulation experiment shows that Q-Min-Min algorithm has the advantage over Min-Min algorithm in the two performance indexes:makespan and cost. It means that the grid resource scheduling algorithm based on multi-QoS requirements in this paper is reasonable and feasible. It can not only save the execution time and cost of resource scheduling, but also satisfy the grid users’multi-QoS requirements. So the paper achieved the purpose of the research.