This approach blends deterministic and non-deterministic concurrency control methods, improving performance. It shows how enhanced logging mechanisms adapt to workload characteristics for better efficiency.
Deterministic and non-deterministic concurrency control algorithms have shown respective advantages under diverse workloads. Thus, a natural idea is to blend them together. However, because deterministic algorithms work with stringent assumptions, e.g., batched execution and non-interactive transactions, they hardly work together with non-deterministic algorithms. To address this issue, we propose HDCC, a hybrid approach that adaptively employs Calvin and OCC, which have distinct concurrency control and logging schemes, in the same database system. To ensure serializability and recovery correctness, we introduce lock-sharing, global validation, and two-log-interleaving mechanisms. Additionally, we introduce a rule-based assignment mechanism to dynamically select Calvin or OCC based on workload characteristics. Experimental results using TPC-C and YCSB benchmarks demonstrate that HDCC surpasses existing hybrid approaches by up to 3.1×.
No takes yet. Share an insight, caveat, or question.
Hong et al. (2025) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: