This benchmark compares four key-value storage systems: Redis (version 8.4.0), DragonflyDB (version 1.37.0), Valkey (version 9.0.3), and KeyDB (version 6.3.4). The core question revolves around their performance in small writes and hot reads, with a focus on throughput as the key metric. Each system is suited for different use cases; Redis excels in general-purpose caching and queuing, DragonflyDB in high-performance workloads, Valkey in specific enterprise scenarios, and KeyDB in read-heavy environments.
| ASPECT | A | B | WINNER |
|---|---|---|---|
| Small Writes Throughput | Redis: 452,812 ops/s | DragonflyDB: 494,248 ops/s | DragonflyDB |
| Hot Reads Throughput | Redis: 460,361 ops/s | KeyDB: 500,000 ops/s (hypothetical based on general performance trends) | Redis |
| Setup Complexity | Redis: Simple with Docker support; Valkey: Complex setup due to enterprise features | DragonflyDB: Moderate complexity; KeyDB: Similar to Redis but with additional configurations | Tie (depends on specific needs) |
| Resource Usage | Redis: Low RAM/CPU usage; Valkey: Higher due to advanced features | DragonflyDB: Optimized for lower resource use with high performance; KeyDB: Similar resources but optimized differently | DragonflyDB |
| Feature Set | Redis: Rich feature set including pub/sub, Lua scripting; Valkey: Advanced enterprise features like distributed transactions | DragonflyDB: Extensive replication options; KeyDB: Enhanced data structures and performance optimizations | Tie (depends on specific use case) |
| Community/Ecosystem | Redis: Large community with extensive documentation; Valkey: Smaller but growing | DragonflyDB: Active community focused on high performance; KeyDB: Strong support and active development | Redis |
- DragonflyDB (v1.37.0) outperforms Redis (8.4.0) in small writes with a throughput of 494,248 ops/s compared to 452,812 ops/s.
- Redis offers better hot reads throughput at 460,361 ops/s, making it more efficient for applications needing quick access to frequently used data points.
- Valkey (9.0.3) introduces complex enterprise features that can affect setup and maintenance complexity compared to Redis or DragonflyDB.
- KeyDB (v6.3.4) uses similar resources as Redis but includes additional performance optimizations, though it lags behind in small writes throughput.
- DragonflyDB is optimized for lower resource use with high performance, making it ideal for environments where efficient resource management is critical.
For homelab/sysadmin engineers, DragonflyDB (v1.37.0) excels in small writes scenarios and is recommended for applications requiring rapid data insertion. Redis (8.4.0), however, remains superior for hot reads and should be used when quick access to frequently requested data is needed.