Benchmark-driven
development
BrewFS treats RandRW, metadata-heavy access, multi-agent scaling, and object-backend performance as first-class benchmark scenarios — measured under published configurations, never as universal claims.
RandRW Benchmark
In the current Redis + RustFS benchmark, BrewFS reaches 92% faster fully drained mixed random-I/O throughput versus JuiceFS — the access pattern that dominates multi-agent workspaces.
Metadata-heavy Benchmark
Directory scans, stat, lookup, rename, and file-attribute operations are constant in agent workspaces. This benchmark isolates the metadata path across pluggable backends.
Multi-agent Scaling
Measure how BrewFS behaves with 1, 2, 4, 8, and 16 concurrent agents sharing the same workspace.
Object Backend Benchmark
Measure object IO performance across LocalFS, MinIO, RustFS, Ceph RGW, AWS S3, and other S3-compatible backends.
Benchmark principles
BrewFS benchmarks always publish workload configuration, backend setup, cache state, mount options, hardware, software versions, and commit hashes. Performance claims must be tied to specific workloads.
- Reproducible. Configurations and commit hashes are published.
- Transparent. Hardware, software versions, and mount options included.
- Workload-specific. Each claim is tied to a defined workload.
- Configuration-aware. Cache state and backend are always reported.
- No misleading universal claims. We never claim BrewFS is faster everywhere.