Performance Monitoring
Sites rarely fail all at once — they get slow first. Performance monitoring tracks your response-time trends over time and alerts on regressions, so a deploy that quietly doubled your latency gets flagged before users start leaving.
How it works
On each run the checker fetches your target and records a full timing breakdown, not just a single number:
- DNS lookup time
- TCP connect and TLS handshake time
- Time to first byte (TTFB)
- Total response time
These are stored per region and charted so you can see p50/p95 trends, compare regions, and correlate a jump with a deploy. Checks run from US-East and additional regions as often as every 30 seconds.
Performance monitoring evaluates against a rolling baseline as well as fixed thresholds — so it can catch a gradual creep that no single static limit would trip.
What triggers an alert
- Total response time (or TTFB) exceeds a fixed warning/critical threshold you set.
- A regression versus the rolling baseline — e.g. p95 latency is materially worse than the trailing window.
- Degradation confirmed across multiple regions, ruling out a single noisy path.
Setting it up
- Add monitor and choose Performance.
- Enter the target URL.
- Set the check interval and regions.
- Configure thresholds (warning + critical) and, optionally, regression sensitivity against the baseline.
- Attach notifications.
Performance and Uptime complement each other: uptime tells you it's down, performance tells you it's getting worse. Many teams run both on their critical endpoints.
Related
- Uptime · Lighthouse · Health Checks
- Notifications
- Marketing: Performance monitoring feature