Grid Resource Management Model Based On Multiple QoS Objectives

Grid technology is a hot technique developed in recent years; it makes full use of all kinds of geographically separated resources to form a super computer. Before technology of Internet carry out the contact among hardware; technology of Web carry out the contact between web pages; in future, technology of grid will help the share of all grid resources in Internal to come true. The resource management is one of core technology of the grid middleware realization, and it includes controlling grid resource and grid service to provide available ability for user, application or other services. So how to design a reasonable and efficient frame and mechanism of grid resource management is one of focuses. However, the work of gird research is focus on solving the heterogeneous system problem, and how to improving the capability of resource management with QoS (Quality of Service) in the different phrase of grid resource lifecycle is less researched.The paper firstly introduce the conception of grid, summarize the status quo and development trend of grid in recent years, and expatiate the importance of the resource management in the research of grid computing. Then thorough analysis and comparisons with regard to characteristic and difference of various models of grid resource management, this paper gives a lifecycle-based model of grid resource management for grid service resources in global scales, which includes grid resource promulgation, discovery and scheduling. Then with the technology of QoS, we discuss in detail the structure design of some QoS and the function relationship between QoS and user's satisfaction in different phases of lifecycle. Two improved algorithm based on multiple QoS objectives, MQ-MinMin and MQ-GA algorithm, can be used to schedule the resource. In these algorithms, the some factors of QoS such as finish time, cost and reliability, are all considered sufficiently, so it can satisfy user's requirement better.At last we simulate our model with the aid of GridSim toolkit and verify that our model is reasonable and efficient.