الفهرس | Only 14 pages are availabe for public view |
Abstract In a distributed database environment, dataisreplicatedtoachievehighavailability. However,complex andexpensivesynchronization protocols are needed to maintain the replica. In general, the cost ofexecutingoperations in a replicated database systemishigher than other forms ofdistributeddatabases (centralization, decentralization, and partitioning), since multiple copies of anobjectmayneedtobeaccessedformaintaining the mutual consistency of the copies. In this work, both CPU-TIME-COSTandoverhead (message cost)arecomputed quantitativelyviaasimulating model for replicateddatabase environment using different replication algorithms, to present a comparative study for these algorithms. CPU- TIME-COST and overhead for these algorithms are compared by running the model under various operational conditions. Finally, a modification in both Quorum consensus and Pessimistic algorithms is introduced to reduce the high cost of exchanged messages. The model is used to demonstrate the change in overhead for both Quorum consensus and pessimistic algorithms before and after modification. 11 |