Why is my app slow to start, and how does WireDoctor help?

[ddsha441981]

Our service boot time crept from 15s to 40s over the last year. Actuator startup JSON is unreadable. Can WireDoctor tell me where the time actually goes?

[ddsha441981]

This is exactly the gap WireDoctor was built for — Actuator gives you raw StartupStep JSON, but nobody ships the analysis layer on top. WireDoctor reads real per-bean instantiation times from BufferingApplicationStartup (no reflection heuristics) and gives you:

1. Ranked slow beans — the Timing tab ranks beans by instantiation cost; wiredoctor.slow-bean-threshold-ms (default 100) flags the slow ones in the report and console.
2. Startup critical path — the instantiation-weighted dependency chain your readiness time actually sits on. Optimizing a bean OFF the critical path buys you nothing; WireDoctor shows you which chain matters.
3. @Lazy payoff ranking — tells you which @Lazy would pay off most instead of leaving you to guess.

And to keep the 15s→40s creep from happening again, arm the performance gates in CI:

wiredoctor.baseline=wiredoctor-baseline.json
wiredoctor.fail-on=startup-time,slow-bean

The startup-time gate is dual-threshold and noise-tolerant, and the slow-bean gate has jitter-margin protection — so normal CI variance doesn't trip it, but a real regression does:

WireDoctorRegressionException: gate 'startup-time' tripped:
  startup time increased by 734ms (18.2%) vs baseline (4025ms -> 4759ms)

Guide: Performance Gates.