StageWise: Accelerating Persistent Key-Value Stores by Thread Model Redesigning

With the emergence of fast NVMe SSDs, key-value stores are becoming more CPU-efficient in order to reap their bandwidth. However, current CPU-optimized key-value stores adopt suboptimal intra- and inter-thread models, hence incurring memory-level stalling and load imbalance that hinder cores from realizing their full potential. <p xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">We present S<small>TAGE</small>W<small>ISE</small>, an CPU-efficient key-value store on fast NVMe SSDs with high throughput. To achieve this, we introduce a new thread model for StageWise to process KV requests. Specifically, S<small>TAGE</small>W<small>ISE</small> converts the processing of each KV request into multiple asynchronous stages, and thus enables pipelining across all stages. S<small>TAGE</small>W<small>ISE</small> further introduces a client-driven share-index architecture to ease inter-thread load imbalance and maximize the pipelining opportunity. Guided by Little’s Law, S<small>TAGE</small>W<small>ISE</small> improves concurrency, and therefore efficiently uses CPU to reach higher throughput. Extensive experimental results show that S<small>TAGE</small>W<small>ISE</small> outperforms CPUoptimized key-value stores (e.g., KVell) by up to 3.5× with writeintensive workloads, and storage-optimized ones (e.g., RocksDB) by over an order of magnitude. S<small>TAGE</small>W<small>ISE</small> also shows higher read performance and excellent scalability under various workloads.