University of Calgary
The authors present a framework for concurrency control and availability in multi-datacenter datastores. While, they consider Google's Megastore as their motivating example, they define general abstractions for key components, making their solution extensible to any system that satisfies the abstraction properties. They first develop and analyze a transaction management and replication protocol based on a straightforward implementation of the Paxos algorithm. Their investigation reveals that this protocol acts as a concurrency prevention mechanism rather than a concurrency control mechanism. They then propose an enhanced protocol called Paxos with Combination and Promotion (Paxos-CP) that provides true transaction concurrency while requiring the same per instance message complexity as the basic Paxos protocol.