On the Design of Adaptive and Speculative Concurrency Control in Distributed Databases

Efficient online transaction processing is key to many database applications, and existing concurrency control protocols perform remarkably well under specific workloads or access patterns that they have been designed for. However, they often do not scale well when the workload is dynamic. To tackle the challenge of dynamic workloads, we propose an Adaptive and Speculative Optimistic Concurrency Control (ASOCC) protocol for effective transaction processing. Based on real-time monitoring of data access frequency, ASOCC adaptively embeds 2PL into the OCC scheme to facilitate superior contention resolution with reduced transaction aborts. Further, ASOCC dynamically inspects the correlation of data accesses and exploits such information to perform speculative transaction restart to save CPU cycles wasted on the processing of transactions that are destined to abort.