SplinterDB: Closing the bandwidth gap for NVMe key-value stores

Modern NVMe solid state drives offer significantly higher bandwidth and lower latency than prior storage devices. Current key-value stores struggle to fully utilize the bandwidth of such devices. This paper presents SplinterDB, a new key-value store explicitly designed for NVMe solid-state-drives. SplinterDB is designed around a novel data structure (the STB^ε-tree) that exposes I/O and CPU concurrency and reduces write amplification without sacrificing query performance. STB^ε-tree combines ideas from log-structured merge trees and B^ε-trees to reduce write amplification and CPU costs of compaction. The SplinterDB memtable and cache are designed to be highly concurrent and to reduce cache misses. We evaluate SplinterDB on a number of micro- and macro-benchmarks, and show that SplinterDB outperforms RocksDB, a state-of-the-art key-value store, by a factor of 6-10× on insertions and 2-2.6× on point queries, while matching RocksDB on small range queries. Furthermore, SplinterDB reduces write amplification by 2× compared to RocksDB.