Fault Tolerance in Distributed Database Management Systems - Improving reliability with RAID

This paper uses the concepts of RAID-5 architecture to enhance fault tolerance in Distributed Database Management Systems. The use of parity bits for recovery from a single database site failure in accordance with the RAID-5 architecture is first demonstrated. The paper then addresses the concerns with this architecture w.r.t. a double disk failure and low availability. The proposed solution aims at providing the DDBMS with high recoverablilty in cases of two database sites failing simultaneously. It also enables the DDBMS to better respond to incoming requests directed at surviving and the failed sites without breaking down and to assist simultaneously, in the reconstruction of the failed site(s) in cases of both single and double database site failures. The proposed solution reduces the cases of complete system failure during a double database site crash significantly.