How monitoring actually works
What each probe proves, why alerts fire when nothing is wrong, and the difference between a service being up and a service working. Written from incidents this site has had.
The site loads, but checkout, login, images or the API are broken
Partial outages are the most common kind and the least likely to be detected. Why a homepage check misses them, and what to check instead.
The anatomy of a false-positive uptime alert
A real one from this site: a page reading 100% uptime and HTTP 200 while the outage feed said the same service had been down for 61 hours. What produced it, and what fixed it.
DNS, TCP, TLS, HTTP: what each check actually proves
A monitor that pings is measuring something different from one that fetches a page. What each layer confirms, what it cannot see, and which failures it will miss entirely.
Why status pages lag real user impact
A vendor status page is a human artifact with a publishing process, not a live measurement. Three structural reasons it is behind you, and what to read instead.
How to choose timeout thresholds without masking real outages
Too short and you page on every cold start. Too long and a hung service looks healthy for minutes. Worse, most timeouts measure the wrong thing entirely.
Synthetic monitoring vs uptime monitoring: what each one misses
One tells you a URL answered. The other tells you a journey completed. They fail in opposite directions, and most teams run only the cheaper one.
Latency, availability, reliability: three different words
They get used interchangeably and they measure different things. Conflating them is how a service reports 99.9% uptime while being unusable.
Why multi-region checks still miss outages
Checking from ten places is better than one and still does not prove a service is healthy. What multi-region buys, what it cannot buy, and where the remaining blind spots are.
How CDN caching makes an outage look inconsistent
The same URL loads for you, fails for a colleague, and recovers when you refresh. That is usually not flakiness — it is a cache, and the pattern tells you which one.
How to monitor your website's uptime for free (and what actually matters)
A practical guide to free uptime monitoring: what a good probe looks like, alert thresholds that avoid false alarms, email vs webhook alerts, and status badges.
Multi-method website checks: why a single request isn't enough
One request can't tell a real outage from a bot block or a protocol quirk. Here's what probing with 4 independent methods in parallel reveals, with real examples.