A Fault-Tolerant Distributed Scheme for Grid Information Services

[[abstract]]The Grid Information Service (GIS), mainly used for resource discovery and monitoring, is a key component of grid system. The resource description and specification should be meditated for efficient search and access. In this paper, we propose a distributed system for grid information services, which deploys a number of registry servers at different regions of the world. A new scheme for registering, updating, querying, and deregistering a resource in registry servers is devised. For the purpose of fault-tolerance and load-balance, the meta-data, including description and specification, of each resource can be replicated and disseminated at some registry servers instead of reproducing resource itself for service discovery. In our scheme, the workload on each registry server is balanced and the faults of registry servers can be tolerated. Also, the user could obtain all resource information satisfied with the query conditions even some of registry servers crashed.