Comparative Performance Evaluation of Evolutionary and Heuristic Database Optimization Techniques in Microservices

This study examines the performance impact of Genetic Algorithm (GA)–based query optimization compared with traditional heuristic rule-based methods in the New Student Admission System of the Graduate School at Universitas Negeri Medan (UNIMED), implemented on a microservices architecture. In recent years, the increasing complexity of distributed systems and the growing volume of transactional data have highlighted the limitations of static heuristic approaches in query optimization, thereby necessitating more adaptive and intelligent techniques. The objective of this research is to evaluate whether GA-based optimization can provide significant performance improvements in such dynamic environments. Performance was evaluated using key metrics, including query latency, throughput, resource utilization, and transactional consistency under varying workloads (read-heavy, write-heavy, and mixed) and simulated user loads ranging from 1,000 to 10,000 users. The experimental results demonstrate that GA consistently outperformed heuristic optimization, reducing average query latency by up to 20.5% and increasing throughput by up to 24.2% through adaptive query plan reordering, indexing strategies, and schema partitioning. Although GA introduced moderate increases in CPU (8.0%) and memory usage (10.4%), it improved transactional consistency under high concurrency.