BrewFSBrewFS
BrewFSBrewFS
Get started
Benchmarks OverviewRandRW BenchmarkMetadata-heavy BenchmarkMulti-agent ScalingReproducibilityBenchmark Methodology
Benchmarks

Metadata-heavy Benchmark

Isolating the metadata path across pluggable backends.

Directory scans, stat, lookup, rename, and file-attribute operations are constant in agent workspaces. This benchmark isolates the metadata path across pluggable backends (SQLx, Redis, etcd, TiKV).

Operations measured

readdir · stat · lookup · rename · getattr

Why it matters for agents

Agents repeatedly walk directories, check file attributes, test file existence, and scan project structures — a single agent already generates significant metadata pressure, and multiple agents amplify it.

RandRW Benchmark

July 24, 2026 Redis + RustFS mixed random I/O results versus JuiceFS.

Multi-agent Scaling

How BrewFS behaves with 1 to 16 concurrent agents.

On this page

Operations measuredWhy it matters for agents