Logo
Deploy Now

Categories

Database Key-Value Store Cache Data Infrastructure In-Memory Redis Alternative Performance

Stars

31,019

Forks

1,229

Watchers

167

Developer links

Last Updated: 08-21-2026 07:56 AM (UTC)

DragonflyDB

With over 30,000 GitHub stars and benchmarks showing 25x the throughput of single-threaded Redis, DragonflyDB is a modern in-memory data store that eliminates the need for complex Redis Cluster deployments by fully utilizing every CPU core on a single machine. Its shared-nothing, thread-per-core architecture written in C++ supports over 200 Redis commands and 13 Memcached commands, making it a true drop-in replacement that requires zero application code changes. A single DragonflyDB instance scales vertically from 8GB to 768GB of RAM across up to 64 cores, replacing entire Redis Cluster topologies with one process while maintaining full compatibility with Strings, Hashes, Lists, Sets, Sorted Sets, Streams, JSON, and Bloom Filters. The novel dashtable data structure and cache eviction algorithm achieve higher hit rates than LRU and LFU with zero memory overhead per entry. Forkless point-in-time snapshotting eliminates the memory spikes associated with Redis BGSAVE, while automatic backup scheduling via cron syntax supports both local disk and AWS S3 cloud storage. Primary-replica replication follows the Redis replication protocol up to version 6.2, and Prometheus-compatible metrics at the default port enable Grafana monitoring dashboards out of the box. DragonflyDB also exposes an HTTP admin interface on its main TCP port for operational monitoring. Running on a dedicated VPS on RepoCloud with guaranteed CPU, RAM, and SSD, full root SSH access, and a browser serial console. BSL 1.1 licensed.

DragonflyDB
DragonflyDB

Benefits

  • Replace Redis Clusters with One Instance
  • A single multi-threaded DragonflyDB instance handles workloads that would require an entire Redis Cluster, eliminating node management, resharding complexity, and cross-slot transaction limitations.
  • Zero Code Change Migration
  • Full Redis and Memcached API compatibility with over 200 supported commands means existing applications connect without modification, using standard Redis clients in any programming language.
  • Forkless Snapshots Save Memory
  • Point-in-time snapshotting without forking eliminates the 2x memory spike of Redis BGSAVE, with automated backups via cron scheduling to local disk or AWS S3 cloud storage.
  • 25x Higher Throughput on Same Hardware
  • The shared-nothing, thread-per-core C++ architecture utilizes all available CPU cores simultaneously, delivering dramatically higher throughput and lower tail latency compared to single-threaded Redis.

Features

  • Multi-Threaded Architecture
  • Shared-nothing, thread-per-core design in C++ utilizes all CPU cores simultaneously, scaling vertically from 8GB to 768GB RAM on a single instance without clustering.
  • Redis API Compatibility
  • Supports over 200 Redis commands including Strings, Hashes, Lists, Sets, Sorted Sets, Streams, JSON, and Bloom Filters, plus 13 Memcached protocol commands.
  • Automated Backups
  • Forkless point-in-time snapshots triggered manually via SAVE/BGSAVE or scheduled via cron, stored to local disk or AWS S3 with automatic restore on startup.
  • Replication and Monitoring
  • Primary-replica replication compatible with Redis OSS up to version 6.2, plus Prometheus-compatible metrics endpoint for Grafana monitoring dashboards.
  • Novel Cache Design
  • Dashtable data structure with SIMD optimizations and a cache eviction algorithm achieving higher hit rates than LRU/LFU with zero per-entry memory overhead.