Comprehensive Evaluation and Opportunity Discovery for Deterministic Concurrency Control

Deterministic concurrency control (DCC) guarantees that the same input transactions produce the same serializable result. It offers benefits in both distributed databases and blockchain systems. Dozens of DCC algorithms have emerged in the past decade. However, there is a lack of comprehensive evaluations for them.

To study the performance of existing DCC algorithms and discover further opportunities, we make the following contributions. First, we abstract five essential features from the existing DCC algorithms: generality, speculative mechanism, version strategy, batch strategy, and concurrency mode. Each distinct combination of these features corresponds to a specific algorithm. Second, we implement 13 DCC algorithms and conduct evaluations focused on their features by using 10 workloads, to conclude each feature’s strengths and weaknesses. Thirdly, based on our feature analysis, we discover opportunities for improvement in two existing DCC algorithms, resulting in performance boosts of up to 2.3x and 3.4x.