Date Added: Apr 2010
A backup server needs to be added to each primary server in the network for increasing the fault tolerance in the distributed database. It is also imperative to ensure that the primary and backup servers are interconnected. This paper seeks to come up with solutions to the problem of high cost of backing up distributed databases. It starts off by introducing a method for reducing the number of packets transferred between the primary and backup servers. The paper's focus is on the use of replicated data for decreasing the costs. It explains the fundamentals and identify the steps need to be taken for implementing the new strategy. A reduction in the number of packets transferred is achieved by ensuring that the transactions sent by the primary server are not replicated on any other system. Therefore, no transfer of the transactions on replicated data needs to be made and hence the decrease in the number of packets. The paper then discusses the various requirements for the smooth functioning of this system. Various non-functional requirements for the system have also been studied. It also seeks to make critical evaluation of the system and to draw the solutions to its shortcomings. The paper also details the implementation process.