dragonflydb/dragonfly

Health v2 · maintenance only

Flags: no_license

Adoption not part of the score

31043 stars · 1234 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-29, confidence not recorded

Dragonfly is a modern in-memory data store that is fully API-compatible with Redis and Memcached, built on a multi-threaded, shared-nothing, thread-per-core architecture. It delivers up to 25X higher throughput and better memory efficiency than Redis on a single instance, with no code changes required to adopt.

Use cases

• replace redis with a faster drop-in compatible cache
• scale a caching layer to millions of QPS on one instance
• reduce infrastructure costs for redis or valkey workloads
• run a high-throughput in-memory key-value store
• use redis-compatible pub/sub as a message queue
• store vector embeddings with a redis-compatible API
• avoid redis cluster resharding complexity with vertical scaling

When to choose

• you need Redis API compatibility with much higher throughput and lower tail latency
• your workload is bottlenecked by Redis's single-threaded architecture
• you want to scale a cache vertically on one large node instead of managing a cluster
• memory efficiency and lower infrastructure cost are priorities

When to avoid

• you depend on Redis modules or ecosystem features Dragonfly does not implement
• you need strict single-threaded Redis semantics or persistence behavior identical to Redis
• your dataset is small and Redis performance is already sufficient
• you require an OSI-approved open-source license, as Dragonfly uses a non-standard license

Facets

service · maturity active

database caching message-queue vector-database databases backend self-hosted self-hosted cloud cpp in-memory-database redis-compatible memcached-compatible key-value-store multi-threaded nosql drop-in-replacement high-throughput real-time linux docker kubernetes

Member repositories

For agents

markdown · JSON · MCP: product_card(name="dragonflydb/dragonfly")