Q: A site is either up or down. Why complicate it?

Q: A site is either up or down. Why complicate it?

A: Because the binary view hides the failures that hurt most. Plenty of real incidents are partial: the homepage loads but checkout 500s, the API works but is 8 seconds slow, one region is broken while others are fine. Your simple up/down monitor calls all of that up.

The myth is that availability is one switch. In practice it's per-endpoint, per-region, and per-latency-threshold.

Recommendation: monitor your critical user journeys separately (login, search, checkout), and treat slow as a form of down by setting a response-time threshold that pages, not just a connection check.

Anticipating the follow-up: pick the latency limit from real user tolerance, often 3-5 seconds for a page, much tighter for an API.

Got a question? Drop it in the comments.