🚀KiviDB: A 9x Faster Redis Alternative in Rust
KiviDB: 9x Faster Than Redis, Built in Rust
TL;DR
KiviDB, a new Rust-based in-memory database, outperforms Redis by 9x in GET operations. It supports full-text and vector search, runs on multiple cores, and is open source.
KiviDB, a new in-memory database written in Rust, is making waves with its performance, outpacing Redis by 9x in GET operations. If you're dealing with high-throughput, low-latency requirements, KiviDB could be a game-changer. It supports full-text and vector search, runs every core on a sharded design, and persists to snapshots and an optional append-only file. With 21 million pipelined GETs per second and under half the p99 latency of Dragonfly, KiviDB is a serious contender for Redis users looking for a speed boost.
Key Points
KiviDB serves 21 million pipelined GETs per second, 9× faster than Redis 8 and 2.3× Dragonfly.
KiviDB has full-text and vector search built in, using Redis Stack–style FT.* commands.
KiviDB's p99 latency for GET/SET pipeline is 0.36/0.42 ms, significantly lower than competitors.
KiviDB's vector search queries/s at ef 64 (recall@10) is 59.3k, outperforming Redis Stack.
KiviDB's memory per key is 177 B, efficient for high-throughput, low-latency workloads.
Why It Matters
If you're running high-throughput Redis workloads, KiviDB's 9x faster GET performance could slash your latency. Its full-text and vector search capabilities, plus efficient memory usage, make it a strong candidate for modern data-intensive applications. However, its non-standard port and licensing model might require careful consideration.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.
Join 3,566 builders reading daily.