A Study on the Impact of Data Replication Strategies on Performance in Distributed Database System

Abstract

Data replication is a fundamental mechanism in distributed database systems, supporting fault tolerance, availability, and reliable data access across networked nodes. Cockroach DB is a cloud native, open-source distributed SQL database that relies on replication and consensus protocols to maintain consistency and resilience in clustered environments. This study investigates the performance implications of synchronous and asynchronous replication strategies within a CockroachDB-based distributed system. A simulation-driven experimental framework is employed to evaluate system behavior under controlled conditions. Key performance metrics including transaction latency, throughput, read and write response times, and replication delay are measured while maintaining consistent cluster topology and configuration parameters. The results show that synchronous replication enforces strong consistency by requiring coordinated acknowledgements from replicas before transaction commitment, but this approach incurs higher coordination overhead, leading to increased latency and reduced throughput. Conversely, asynchronous replication improves write responsiveness and overall throughput by decoupling transaction commits from replica synchronization, although it introduces temporary inconsistencies due to replication lag. The findings highlight the inherent trade-off between consistency and performance in distributed database systems. This study provides practical guidance for selecting appropriate replication strategies based on workload characteristics and application consistency requirements, contributing empirical insight to the configuration and optimization of modern distributed data platforms.