Distributed knowledge base fault tolerant system

Fault tolerance is one of the desirable properties of a distributed system. This paper focuses on developing a fault tolerant distributed knowledge base. This knowledge base uses checkpointing for failure recovery. The study proposes a scheme in which checkpoint is saved only on a set of processors. The set will comprise of only those processors which are participating in the distributed transaction. The checkpoint of a processor is saved in rest of the participating processors. This scheme allows processor failure without using any dedicated checkpoint processor. This scheme does not require complete backup of data as XOR operation is used to store checkpoint.