#Benchmark and Performance

#Table of Contents

• Environment
• Current Memory Benchmark
• Memory OSS Comparison
• Redis Direct Benchmark
• HTTP Middleware Benchmark
• Run the Benchmarks
• Reading the Numbers

#Environment

Latest local run:

These numbers are local-machine measurements, not portable product claims.

#Current Memory Benchmark

Command:

$env:BENCH_JSON='1'
$env:BENCH_RUNS='5'
$env:BENCH_ITERATIONS='200000'
$env:BENCH_KEYS='1000'
npm run benchmark:memory

#Memory OSS Comparison

The OSS comparison uses the historical local script in .devcodex/flex-rate-limit/tmp/oss-rate-limit-bench-20260531/compare-oss.cjs.

#100,000 checks / 1,000 keys

#200,000 checks / 5,000 keys

#100,000 checks / 1 hot key

#Redis Direct Benchmark

Command:

$env:BENCH_JSON='1'
$env:BENCH_ITERATIONS='5000'
$env:BENCH_KEYS='500'
$env:BENCH_CONCURRENCY='1,32'
npm run benchmark:redis

#HTTP Middleware Benchmark

Command:

$env:BENCH_JSON='1'
$env:BENCH_DURATION='5'
$env:BENCH_CONNECTIONS='50'
$env:BENCH_KEYS='500'
npm run benchmark:http

#Run the Benchmarks

npm run benchmark:memory
npm run benchmark:redis
npm run benchmark:http

Set BENCH_JSON=1 for machine-readable output. Redis benchmarks skip cleanly if Redis is unavailable.

#Reading the Numbers

• Use Memory results for single-process direct check() hot paths.
• Use Redis direct results for distributed counter overhead.
• Use HTTP results for Express-style middleware overhead.
• Do not compare these numbers to production traffic without matching Node.js, CPU, Redis topology, concurrency, key distribution, and application work.