🩺 Vitals
- 🟢 Last active: 2026-10-08
- 📦 Latest release: v2.0.2 (2026-10-07)
- 🐞 Open issues: 328
- 🌟 Stars: 31.8k
What do these metrics mean?
- Last active: when code was last pushed, as of our last check. The dot is green when that was recent, grey otherwise. A long gap can mean a tool is finished and stable, not only unmaintained.
- Latest release: the most recent tagged, packaged version the maintainers published. Not every healthy project tags releases.
- Open issues: unresolved reports and requests. A high number is normal for a popular project and is not a warning on its own.
- Stars: how many people bookmarked the project on its forge. A rough popularity signal, not a measure of quality.
🏗️ Profile
- Official: dragonflydb.io
- Source: github.com/dragonflydb/dragonfly
- License: BSL 1.1 (No DBaaS resale; Apache 2.0 after 4 yrs)
- Deployment: Docker | Kubernetes | Native Binary
- Data Model: In-Memory Key-Value (Redis and Memcached protocols) / RDB snapshots
- Jurisdiction: Israel 🇮🇱 (DragonflyDB Ltd.)
- Compliance (SaaS): SOC 2 Type II (Dragonfly Cloud)
- Compliance (Self-Hosted): Self-Hosted (User Managed)
- Complexity: Low (2/5) - Single multi-threaded binary; Docker image and Kubernetes operator provided
- Maintenance: Low (2/5) - Binary updates and periodic RDB snapshots; optional replication for read-scaling and failover, and the vertical model avoids Redis cluster ops
- User Management: Moderate (3/5) - Redis-compatible ACLs and native TLS; SSO, MFA, and bring-your-own-cloud (BYOC) are Dragonfly Cloud paid-tier only
1. The Executive Summary
What is it? Dragonfly is an in-memory datastore that speaks the Redis and Memcached wire protocols, so existing clients connect without code changes. Its design goal is vertical scaling: a multi-threaded architecture that uses every core on a single large machine, where Redis runs effectively single-threaded and reaches for clustering to grow. It is built and licensed by DragonflyDB Ltd., a commercially-backed company in Israel, under the Business Source License (BSL 1.1), which makes it source-available rather than open source. The full engine self-hosts via Docker, a Kubernetes operator, or a single binary, and exports point-in-time snapshots in Redis .rdb format for a clean exit back to any Redis-compatible store. A managed Dragonfly Cloud exists for teams that prefer not to run it themselves.
The Strategic Verdict:
- 🔴 For teams that need an open-source licence, or that build a hosted data service: Caution. Dragonfly is BSL, not open source, and its licence specifically forbids offering it to third parties as an in-memory data service. If your product is a managed cache or DBaaS, this is a hard blocker. If an open licence is the only requirement, Valkey (BSD, Linux Foundation) or Redis under its AGPL option are the on-mission choices.
- 🟢 For teams hitting the ceiling of a single Redis node: Strong Buy. The multi-threaded engine makes vertical scaling a real option, often replacing a Redis cluster with one large instance. That removes shard management and cluster-failover choreography while keeping drop-in client compatibility and a snapshot-based exit path.
2. The "Hidden" Costs (TCO Analysis)
| Cost Component | AWS ElastiCache (SaaS) | Dragonfly (Self-Hosted) |
|---|---|---|
| Licensing | Per-node-hour rental, billed idle or busy | $0 (BSL, internal production use) |
| Scaling Model | Horizontal: add nodes and shards, each billed | Vertical: one node uses all cores |
| Data Residency | AWS region | 100% self-owned |
| Data Exit | Engine-dependent export | Redis .rdb snapshots (portable) |
3. The "Day 2" Reality Check
🚀 Deployment & Operations
- Installation: A single multi-threaded binary, shipped as a Docker image, a Kubernetes operator for high availability and failover, or a native binary. Redis and Memcached protocol compatibility means existing clients, libraries, and monitoring tools connect unchanged.
- Scalability: Vertical-first. One process saturates all CPU cores and large memory on a single node, which is the core design claim: replace a multi-node Redis cluster with a single instance and drop the shard-and-cluster management overhead. Replication supports read scaling and failover.
🛡️ Security & Governance (Risk Assessment)
- Jurisdiction & Geopolitics (Israel 🇮🇱): Dragonfly is developed by DragonflyDB Ltd., incorporated in Israel, with terms governed by Israeli law and disputes venued in Tel Aviv. Israel is a close Western technology partner but sits outside both the EU GDPR perimeter and the US legal system, so its data-processing framework stands on its own. For self-hosted deployments this is largely moot, because no application data reaches the vendor. The one outbound signal is a daily version check to Dragonfly's servers, on by default and switched off with a single flag (
--version_check=false) for air-gapped operation. The vendor is not directly subject to the US CLOUD Act unless you engage the managed Dragonfly Cloud through a US entity. - The Compliance Shift: Dragonfly Cloud, the managed offering, holds a SOC 2 Type II attestation. That certificate covers the vendor's cloud operations only and does not transfer to a binary you run yourself. Self-hosting inverts the shared-responsibility model: the software provides the datastore, native TLS, and Redis-compatible ACLs, and you own everything above that. Meeting SOC 2 or HIPAA on your own infrastructure means configuring TLS, enforcing ACLs, and adding data-at-rest encryption and backup controls at the platform layer. The engine supports the controls; the accountability is yours.
- License Risk (the BSL trap): Dragonfly uses the Business Source License 1.1. Internal production use is free, including commercial use inside your own product. The restriction is specific: the Additional Use Grant forbids offering Dragonfly to third parties as an in-memory data store service, so a managed cache or DBaaS built on it needs a commercial licence. Each released version converts to Apache 2.0 four years after its own publication, so older versions open up over time while anything recent always carries the restriction. For internal caching and application state, the standard enterprise pattern, the licence imposes no cost.
4. Market Landscape
🏢 Proprietary Incumbents
- AWS ElastiCache: The default managed cache for AWS shops, now offering Redis, Valkey, and Memcached engines. Teams weigh Dragonfly to escape per-node rental billing and a horizontal-only scaling model.
- Redis Cloud: The vendor's managed and enterprise tier. Evaluated against a self-hosted Dragonfly when the goal is to drop the managed-service premium while keeping Redis compatibility.
🤝 Open Source Ecosystem
- Valkey: The BSD-licensed, Linux Foundation fork of Redis created after the 2024 relicense, and the community-governed open-source default for teams whose first requirement is an open licence rather than Dragonfly's vertical-scaling performance.
- Redis: The original, now offered under a three-way licence choice that includes OSI-approved AGPLv3; self-hostable, and the compatibility baseline Dragonfly targets.