DEV Community

Alex Spinov
Alex Spinov

Posted on

DragonflyDB Has a Free Redis Alternative That Uses All Your CPU Cores

#dragonflydb #redis #database #performance

Redis is single-threaded. DragonflyDB is multi-threaded from the ground up — handling 25x more throughput on the same hardware while being API-compatible.

The Numbers

Redis 7 DragonflyDB
Architecture Single-threaded Multi-threaded
Max throughput ~1M ops/s (1 core) ~25M ops/s (multi-core)
Memory efficiency 1x 2-5x better (shared nothing)
Snapshotting Fork (2x RAM spike) Incremental (no spike)
Max dataset size Limited by 1 thread Limited by total RAM
API compatible — 99%+ Redis/Memcached

Setup 

docker run --ulimit memlock=-1 -p 6379:6379 docker.dragonflydb.io/dragonflydb/dragonfly
Enter fullscreen mode Exit fullscreen mode

Use your existing Redis client — zero code changes.

import Redis from "ioredis";

// Connects to DragonflyDB exactly like Redis
const redis = new Redis(6379, "localhost");

await redis.set("key", "value");
const val = await redis.get("key");

// All Redis commands work
await redis.hset("user:1", { name: "Alice", email: "alice@test.com" });
await redis.lpush("queue", "task1", "task2", "task3");
await redis.zadd("leaderboard", 100, "player1", 200, "player2");
Enter fullscreen mode Exit fullscreen mode

Why Multi-Threading Matters

Redis uses 1 CPU core. On a modern 16-core server, you waste 15 cores or run 16 Redis instances (nightmare to manage).

DragonflyDB uses all cores automatically:

16-core server:
  Redis: 1 instance per core × 16 = complex, shared-nothing, manual sharding
  Dragonfly: 1 instance, uses all 16 cores = simple, automatic
Enter fullscreen mode Exit fullscreen mode

Memory Efficiency

DragonflyDB uses a novel memory allocation strategy:

  • No fork for snapshots — Redis forks the process for RDB saves, temporarily doubling RAM usage. Dragonfly uses incremental snapshots.
  • Dash data structure — more compact than Redis's hash tables
  • Result: Same data, 30-60% less RAM

Supported Commands

  • All string commands (GET, SET, MGET, INCR, etc.)
  • All hash commands (HSET, HGET, HGETALL)
  • All list commands (LPUSH, RPOP, LRANGE)
  • All set commands (SADD, SMEMBERS, SINTER)
  • All sorted set commands (ZADD, ZRANGE, ZRANGEBYSCORE)
  • Pub/Sub
  • Streams
  • Lua scripting
  • Transactions (MULTI/EXEC)
  • Cluster mode (emulated)

Configuration 

docker run -p 6379:6379 docker.dragonflydb.io/dragonflydb/dragonfly \
  --maxmemory 4gb \
  --dbfilename dump \
  --dir /data \
  --proactor_threads 8
Enter fullscreen mode Exit fullscreen mode

When to Use DragonflyDB

Choose DragonflyDB when:

  • You need more throughput than single-threaded Redis
  • Memory efficiency matters (large datasets)
  • You want simpler ops (1 instance vs. Redis Cluster)

Stick with Redis/Valkey when:

  • You need Redis Cluster (native, not emulated)
  • Redis Modules (RediSearch, RedisJSON) are critical
  • You want longest track record

Need high-performance data infrastructure? I build caching solutions and data tools. Email spinov001@gmail.com or check my Apify tools.

Top comments (0)