Gomap is a low-level, memory-mapped storage engine designed for high-performance workloads where read throughput and latency matter. Inspired by Redis and engineered as an alternative to LSM-based key-value stores like BadgerDB, Gomap uses custom on-disk hashmaps to deliver blazing-fast GET operations at scale.
“Hashmaps are to reads what LSM trees are to writes — but without the compaction overhead.”
- Provide a Redis-style server interface backed by Gomap.
- Enable side-by-side benchmarking against BadgerDB.
- Showcase performance under large key counts (tested up to 50 million keys).
- Demonstrate low-latency, high-throughput read performance using mmap-backed hashmaps.
- Visualize comparative performance via native Go + Matplotlib plots.
- Redis protocol compatibility via
redcon. SET/GETcommands mapped to Gomap or Badger engines.- Clean benchmark suite with CSV export and performance plots.
- Automated benchmarking via
Makefile. - Scales efficiently with large key volumes.
Gomap significantly outperforms BadgerDB in high-read scenarios — especially as key volume increases.
See this chart from our benchmark suite:
| Engine | Keys | SET RPS | SET p50 | GET RPS | GET p50 |
|---|---|---|---|---|---|
| Gomap | 1,000 | 336,000 | 1.98 ms | 1,008,000 | 0.47 ms |
| Gomap | 10,000 | 434,782 | 5.03 ms | 1,666,667 | 0.89 ms |
| Gomap | 1,000,000 | 472,367 | 4.34 ms | 1,721,170 | 0.91 ms |
| Gomap | 50M | 468,872 | 4.23 ms | 1,673,304 | 0.94 ms |
| Badger | 1,000 | 251,999 | 3.39 ms | 1,008,000 | 0.52 ms |
| Badger | 1,000,000 | 410,004 | 7.55 ms | 1,168,224 | 0.10 ms |
| Badger | 50M | 425,659 | 7.23 ms | 1,057,373 | 0.09 ms |
✨ Gomap consistently delivers 30–50% higher GET throughput and 30–60% lower latency than Badger at scale.
gomap/
├── cmd/
│ └── benchmarkmain/ # Entry point for benchmark suite
├── redisserver/
│ ├── gomapredis/ # Redis server wrapper for Gomap
│ └── badgerredis/ # Redis server wrapper for Badger
├── benchmark/
│ ├── config.go # CLI flag parsing
│ ├── runner.go # Benchmark orchestration
│ ├── report.go # CSV + terminal output
│ └── plot.go # Native Go plot support (optional)
├── benchmark/benchmark_performance_combined.png
├── go.mod
└── ...