A FAULT-TOLERANT MODEL FOR REPLICATION IN DISTRIBUTED-FILE SYSTEMS

In this paper, we propose a new fault-tolerant model for replication in distributed-file systems. We combine the advantages of the modular redundancy, primary-stand-by, and weighted priority schemes to address the fault-tolerant model. To fulfill the fault-tolerant requirements, we use the ideas of directoryoriented replication and the extended prefix table while incorporating a two-level heap data structure. In a consideration of practical circumstances, the heap, using an array instead of a linked-list tree structure, has a dynamically adjustable property, which can be used to easily modify the tree structure in the distributed-file system. Therefore, this fault-tolerant model provides low-cost overhead, a variety of transparency capacities, a fault-tolerant capacity, and a parallel commitment capacity on backup replicas for distributed-file systems.