Adaptive Query-Driven Fragmentation for Performance Optimization in Distributed Databases

Abstract

Distributed databases improve scalability and availability but introduce challenges related to query performance and data management. Fragmentation is a key technique used to enhance efficiency by reducing irrelevant data access. This paper proposes an adaptive fragmentation approach based on real query patterns to optimize distributed database performance. The method applies horizontal and derived fragmentation techniques driven by frequently executed queries, followed by a genetic algorithm to reduce redundant predicates and minimize fragment growth. The approach was implemented and evaluated using the TPC-H benchmark on a PostgreSQL-based distributed environment. Experimental results demonstrate notable reductions in query response time and communication overhead while maintaining high query coverage. The findings confirm that query-driven adaptive fragmentation can effectively improve distributed database performance and resource utilization.