How to Run Dragonfly with Docker: A Quick Start Guide

Getting Started with Dragonfly on Docker

Dragonfly is the fastest way to get a high-performance Redis-compatible data store running on your own infrastructure. In this quick start guide, you will have Dragonfly up and running with Docker in under 5 minutes — including vector search for your AI applications.

Prerequisites

Before starting, ensure you have Docker installed on your system. Dragonfly runs as a single lightweight container with minimal overhead.

Step 1: Pull the Image

docker pull dragonflydb/dragonfly:latest

Step 2: Run Dragonfly

Start Dragonfly with default settings. It listens on port 6379 (Redis-compatible):

docker run --name dragonfly -d --network=host dragonflydb/dragonfly:latest

Or, for a more portable setup with explicit port mapping:

docker run --name dragonfly -d -p 6379:6379 dragonflydb/dragonfly:latest

Step 3: Connect and Verify

Use the Redis CLI to connect to your running Dragonfly instance:

docker exec -it dragonfly redis-cli PING

You should see PONG — Dragonfly is live!

Step 4: Enable Vector Search

Dragonfly supports HNSW vector indexes out of the box. Create a vector index and add embeddings:

# Create an HNSW index for 768-dimensional vectors
FT.CREATE idx ON HASH PREFIX 1 doc: SCHEMA embedding VECTOR HNSW 6 DIM 768 TYPE FLOAT32 DISTANCE_METRIC COSINE

# Add a document with an embedding
HSET doc:1 embedding "0.1,0.2,...0.768"

# Search by similarity
FT.SEARCH idx "*=>[KNN 10 @embedding $vec]" PARAMS 2 vec "0.1,0.2,..." DIALECT 2

Docker Compose Setup

For production deployments with persistence and health checks:

version: '3.8'
services:
  dragonfly:
    image: dragonflydb/dragonfly:latest
    restart: unless-stopped
    ports:
      - "6379:6379"
    volumes:
      - ./data/dragonfly:/data
    healthcheck:
      test: ["CMD", "redis-cli", "PING"]
      interval: 10s
      timeout: 3s

Performance Tuning

Dragonfly automatically uses all available CPU cores and memory. For fine-tuning, pass flags:

docker run --name dragonfly -d \
  -p 6379:6379 \
  -v $(pwd)/data:/data \
  dragonflydb/dragonfly:latest \
  --proactor_threads=4 --cache_mode=true

Conclusion

You now have Dragonfly running as a high-performance in-memory data store with vector search capabilities. Whether you are using it as a drop-in Redis replacement, a semantic cache for LLMs, or a vector database for RAG, Dragonfly delivers exceptional performance with minimal setup effort.