Reconciling Transactional and Non-Transactional Operations in Distributed Key-Value Stores

NoSQL databases were initially designed to provide extreme scalability and availability for Internet applications, often at the expense of data consistency. The recent generation of Web-scale databases fills this gap, by offering transaction support. However, transaction processing implies a significant performance overhead on online applications that only require atomic reads and writes. The state-of-the-art solutions are either static separation of the data accessed by transaction-enabled and native applications, or complete "transactification" of the latter, which are both inadequate.

We present a scalable transaction processor, Mediator, that enjoys the best of both worlds. It preserves the latencies of atomic reads and writes, without compromising data safety. We introduce a lightweight synchronization protocol that enables conflict resolution between transactions and native operations that share data in a distributed database. We evaluate Mediator's implementation on top of the HBase key-value store on a large-scale testbed, and show that it substantially outperforms the traditional approach on a vast majority of mixed workloads. In particular, Mediator achieves a significantly larger throughput for all workloads in which the fraction of native operations exceeds 50%.