Cache Engines
Dragonfly#
Ultra-fast, multi-threaded, Redis-compatible.
- Up to 25× throughput versus single-threaded engines
- Shared-nothing architecture that scales with CPU cores
- Fully Redis-compatible (RESP2/3)
- Built-in tiered storage (RAM + SSD) — Dragonfly Tier
- Automatic memory management with a dashtable allocator
Best for high-throughput workloads where raw performance matters most. Used on Speed, Business, and Enterprise tiers.
Valkey#
Open-source, cluster-ready, community-driven.
- True Cluster Mode with automatic sharding
- Linear horizontal scaling across nodes
- Automatic failover with replica promotion
- 100% Redis protocol compatible
- Active open-source community (Linux Foundation)
Best for cluster deployments that need horizontal scaling and high availability. Used on Free, Dev, and Pro tiers.
How to choose an engine#
| Feature | Dragonfly | Valkey |
|---|---|---|
| Max throughput | Highest | High |
| Cluster mode | — | Yes |
| Tiered storage | Yes (SSD) | — |
| Multi-threaded | Native | io-threads |
| Redis compatibility | RESP2/3 | RESP2/3 |
| License | BSL 1.1 | BSD-3 |
Setting the engine via API#
POST /api/v1/instances
{
"name": "my-cache",
"engine": "dragonfly", // "dragonfly" | "valkey"
"tier": "hobby"
}
See the API Reference for all available fields.
Related#
If your workload needs multi-node sharding and automatic failover, read Cluster Mode next — it runs on the Valkey engine described here. For provisioning engines as code, see the Terraform Provider.