Logo
Deploy Now

Categories

Database Key-Value Store NoSQL Storage Redis C++ Apache

Stars

4,401

Forks

652

Watchers

82

Developer links

Last Updated: 08-21-2026 01:42 PM (UTC)

Kvrocks

Every Redis client you already use connects to Kvrocks without a single code change, but instead of holding your entire dataset in RAM, data lives on SSD through RocksDB, turning terabytes of memory cost into pennies of disk. An Apache Software Foundation top-level project, Kvrocks supports strings, hashes, lists, sets, sorted sets, streams, bitmaps, JSON documents, TimeSeries data points, Bloom filters, Cuckoo filters, and HyperLogLog structures, all persisted to disk with in-memory caching for hot data access. Asynchronous replication using binlog similar to MySQL provides data durability across replicas, while Redis Sentinel integration enables automatic failover when master or replica nodes fail. The proxyless centralized cluster architecture distributes data across shards while remaining fully compatible with standard Redis cluster SDKs and clients. Token-based namespaces provide multi-tenant isolation with authentication per namespace, going beyond Redis SELECT's numbered database model. RocksDB's LSM-tree storage engine provides efficient compression through configurable compaction strategies, reducing disk footprint dramatically while maintaining sub-millisecond reads for cached keys. Migration tooling includes RedisShake for Redis-to-Kvrocks live migration and kvrocks2redis for reverse migration, enabling gradual adoption without service interruption. The kvrocks_exporter exposes Prometheus-compatible metrics for monitoring, and OpenTelemetry integration provides distributed tracing. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. Apache 2.0 licensed.

Kvrocks
Kvrocks
Kvrocks
Kvrocks
Kvrocks

Benefits

  • Drop-In Redis Replacement on SSD
  • Full Redis protocol compatibility allows any existing Redis client library to connect without code changes while RocksDB stores data on SSD instead of memory, dramatically reducing infrastructure costs.
  • Distributed Cluster with Redis Compatibility
  • Proxyless centralized cluster architecture distributes data across shards while remaining compatible with standard Redis cluster SDKs, enabling horizontal scaling with familiar client-side tooling.
  • Async Replication and Sentinel HA
  • Binlog-based asynchronous replication similar to MySQL provides data durability across replicas, with Redis Sentinel integration enabling automatic failover when master or replica nodes fail.
  • Complete Redis Data Structure Support
  • Supports strings, hashes, lists, sets, sorted sets, streams, bitmaps, JSON, TimeSeries, Bloom filters, Cuckoo filters, and HyperLogLog with RocksDB disk-backed persistence and in-memory hot data caching.

Features

  • RocksDB Storage Engine
  • Facebook's LSM-tree storage engine provides efficient disk-based persistence with configurable compaction, compression, and column family isolation for metadata and data.
  • Token-Based Namespaces
  • Multi-tenant namespace isolation with per-namespace authentication tokens provides logical database separation beyond Redis SELECT, with independent access control per tenant.
  • Migration Tooling
  • RedisShake enables live migration from Redis to Kvrocks, while kvrocks2redis supports reverse migration for gradual adoption without service interruption.
  • Prometheus Metrics Export
  • The kvrocks_exporter exposes server statistics, replication lag, command throughput, and RocksDB compaction metrics in Prometheus-compatible format for monitoring dashboards.
  • Stream and JSON Support
  • Redis Streams with consumer groups and pending entry lists plus JSON document storage with path-based access enable event sourcing and document workflows on disk.