RandRW Benchmark
July 24, 2026 Redis + RustFS mixed random I/O results versus JuiceFS.
In the July 24, 2026 Redis + RustFS benchmark snapshot, BrewFS records 92% more fully drained mixed random-I/O throughput versus JuiceFS (218.60 vs 113.86 MiB/s), and 51% more foreground mixed random I/O (495.54 vs 328.76 MiB/s).
| Metric | BrewFS | JuiceFS | Advantage |
|---|---|---|---|
| Mixed random I/O, foreground | 495.54 MiB/s | 328.76 MiB/s | 51% faster (~1.51×) |
| Mixed random I/O total, fully drained | 218.60 MiB/s | 113.86 MiB/s | 92% faster (~1.92×) |
| Random read | 1,726.10 MiB/s | 1,256.80 MiB/s | 37% faster (~1.37×) |
This result applies to the tested RandRW workload configuration. It should not be interpreted as a universal claim that BrewFS is faster in every scenario. Across the full matched matrix, BrewFS leads 15 of 24 rows; JuiceFS still leads sequential read and strictly drained pure writes.
Test configuration
| Item | Value |
|---|---|
| Benchmark type | RandRW |
| Comparison | BrewFS vs JuiceFS |
| Workload model | Agent-style random read/write |
| Filesystem access | FUSE |
| Per-job size / runtime | 512 MiB / 20 s (time-based) |
| Cache state | Durable prefill, drain, cache-preserving remount |
| Benchmark tool | fio with io_uring |
| Mount profile | writeback-throughput-profile |
Metrics captured: IOPS, throughput, average latency, P95, and P99 — for both cold-cache and warm-cache states.
Recommended fio profile
[agent-randrw]
rw=randrw
rwmixread=70
bs=4m
iodepth=1
numjobs=8
size=512m
runtime=20
time_based=1
direct=0
group_reporting=1Replace these parameters with the real benchmark configuration before publishing your own numbers. See Reproducibility.