Research On Grid Task Scheduling Algorithm Restricted By Multiple Quality Of Service

Grid integrates all kinds of computer resources, and turns them into a widespread available, reliable, and standard computing power. This requires Grid middleware resource management system can provide an effective resource management, use better task scheduling strategy ensures that the quality of service of users, so that resources will be better utilized. Along with the extension of grid study, the quality of service of users is a new factor need to be considered in task scheduling study.A specific task of scheduling algorithm is close to a task scheduling model, which is the base of task scheduling algorithm. Based on studying layer model of grid QoS, a QoS-based task scheduling model is proposed, which is determined by the three models of task, computing platform and performance. Study user's quality of service requirements in depth. Based on measurable and policy of QoS requirements, users` QoS requirements are modeled, a formulation definition of utility function is provided, which is used to represent the gain after user's QoS is met. And utility functions in different service levels of timeliness, reliability, security and data accuracy requirements of a task, are given. Mapping utilities gained from each QoS requirement of a task to a unified benefit affected by priority of the task, preparing for the study of QoS-base scheduling algorithm.Total utility gained by user and the completion time of a task which includes the transmission time of the input and output of a task are Considered, and thought of as scheduling goals, an improved QoS-based heuristic algorithm(MQR-HA) is proposed. In order to apply to the dynamic of the grid environment, think of MQR-HA as a scheduling event, and design a dynamic scheduling algorithm by way of regular time interval.The simulation results show that the MQR-HA can obtain better performance from the total utility, the number of discarded tasks, and makespan and so on.