Spooky includes a dedicated benchmark crate and helper scripts for repeatable performance work.

Components

• crates/bench/ provides the benchmark CLI and report generation
• scripts/bench-micro.sh runs the micro suite
• scripts/bench-macro.sh runs the macro suite
• scripts/bench-gate.sh compares current runs to baseline reports
• scripts/bench-promote-baseline.sh promotes a report into the stored baseline set

What The Benchmark Suite Covers

• route lookup behavior
• load-balancer selection behavior
• connection lookup behavior
• header collection behavior
• macro traffic-mix workloads
• long-lived stream workload models

Why It Matters

This project has a large hot path in the edge runtime. Benchmarking is part of the quality bar for changes that affect:

• routing
• connection lookup
• balancing
• buffering
• stream lifecycle

Operational Use

• use micro benchmarks for algorithmic regressions
• use macro benchmarks for end-to-end hot-path behavior
• use baseline gating for release confidence rather than one-off headline numbers