Recovery mechanism of cooperative process chain in grid

A series of distributed processes usually need to be created in order to complete a user´s task in grid environment, and these processes which have been created at different grid site form a process organization (called process-tree). Due to the dynamic of grid resource and some uncertain factors, some process nodes in this tree may be not accessible by the other related ones. It leads to a process-tree broken problem. As a result, it would block further execution of the processes we have created. Up to now, there is no good solution to solve the problem. In this paper we explore this problem and discuss how to handle it if some inaccessible nodes in process-tree happened, and how to ensure the integrity of organization structure of distributed processes. After making a comparison and analysis with the traditional distributed systems, we provide a new mechanism to increase reliability of grid-based computing environments and restore the inaccessible process-node over process tree in grid environment and endeavoring to keep the integrity of the original process organization structure. This new mechanism would adopts two types of additional information and two extra process actions to help restore the original process but not disturb the existing dynamic characters of grid resources, and grid tasks can be performed efficiently as well.