Title: A Comparative Study of Query Concurrency Control Techniques in Relational Database Management Systems
Authors: BIDUAYA MUKANA Clément
Volume: 10
Issue: 8
Pages: 33-36
Publication Date: 2026/08/28
Abstract:
Concurrency control is a crucial factor determining the performance optimization and reliability of Relational Database Management Systems (RDBMS) [source: 1]. This paper presents a rigorous empirical comparative evaluation between two major isolation algorithms implemented within the MySQL 8.0 InnoDB storage engine: Multi-Version Concurrency Control (MVCC) and Strict Two-Phase Locking (S2PL). Using the Sysbench benchmarking tool, Online Transaction Processing (OLTP) workloads were simulated on an initial dataset of 100,000 records across moderate (10 concurrent threads) and intensive (50 concurrent threads) user environments. Experimental results demonstrate the operational superiority and high scalability of the MVCC protocol, achieving a throughput of 29.48 transactions per second (TPS) under 50 threads with zero deadlocks. Conversely, the S2PL protocol suffered an exponential performance degradation under high data contention, causing critical system saturation marked by 2,005 deadlocks within 30 seconds and a doubled average latency of 3,167.06 ms. This study provides clear engineering and managerial guidelines for information systems, recommending the strict adoption of MVCC for high-concurrency OLTP applications while restricting S2PL to critical financial transactions requiring absolute serializability