摘要

With the rapid advances in grid technology, grid reliability becomes an issue of increasing practical concern and research attention. However, managing grid reliability is not easy due to the complicated, large-scale and dynamic natures. This paper studies the reliability problem of grid service imposed by software or hardware failures, and presents a new approach to analyze grid service reliability subject to failures. The grid service process is described and the service execution is defined as a directed acyclic graph. Then the failure types and contributing factors happened to grid service process are expatiated. Moreover, the reliability evaluation equations about request process, matchmaking process, and service execution are respectively suggested. Lastly, a numerical example is given to illustrate these equations. The results show that our approach is feasible.