Introducing SurrealMX: SurrealDB’s new in-memory storage engine

SurrealDB has introduced SurrealMX 1.0, an embedded in-memory transactional key-value storage engine aimed at low-latency, high-concurrency workloads such as real-time services, caching, simulations, ephemeral application state, testing, and browser-based applications. Built around a concurrent adaptive radix tree, multi-version concurrency control, lock-free reads, and a lock-free commit ring, it supports snapshot and serializable snapshot isolation, rollbacks, savepoints, and lower-overhead direct operations. In company benchmarks on an AMD Threadripper system with 15 million keys and 48 worker threads, SurrealMX reportedly reached more than 21.5 million point reads and 18 million range scans per second, outperforming in-memory configurations of LMDB and RocksDB in the cited tests. Although its fastest mode stores data only in memory and loses it when the process ends, it offers optional append-only logs, snapshots, asynchronous persistence, configurable fsync behavior, and LZ4 compression for recovery needs. SurrealDB says the engine was validated through 25 million deterministic simulated operations with no divergences and positions SurrealMX alongside disk-backed SurrealKV, which is intended for durable persistent workloads.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.