Design of Distributed Voting Replication Model for Replica Control

The objective of replication in data-centric distributed systems is to improve a service's performance, to increase its availability, or to make it fault-tolerant and reliable. In this paper, we present a replication extension model based on quorum-based voting that enforces consistent operations in a distributed system. The model uses a hybrid combing the static and dynamic voting, to configure the trade-off between availability and replica consistency. We contribute with an enhanced voting replication system with fault tolerance in the presence of server and network failures. The performance of our voting replication protocol is evaluated. As a part of the experimental studies, the effects of varying the voting frequencies and synchronization delays versus voting trade-off are examined. The effects of the latency of operations are also studied. Our preliminary findings are presented.