Response-Time Charts: Watching Performance, Not Just Uptime
Jul 29, 2026 · 2 min read
"Up" is not the same as "fine"
A check that returns a 200 in 4 seconds passes. It also tells you almost nothing is wrong yet, and something is about to be. Uptime alone can't see that — it only has two states, and "slow" isn't one of them.
What gets recorded
Every single check — not just the failures — stores its response time in milliseconds. That number feeds a chart on the monitor's detail page, plotted against your check history, so you can see the trend instead of just the current number.
Why the trend matters more than any one number
A single slow check is easy to dismiss. A response time that's crept up over two weeks, even while every check still passes, is a server running low on headroom — a database that needs an index, a process leaking memory, traffic that's outgrown the box it's running on. The chart makes that visible days before it turns into the timeout that finally opens an incident.
Where to find it
Open any monitor's detail page. The response-time chart sits right alongside the incident log, because the two are meant to be read together: the log tells you when something broke, the chart often tells you it was already breaking.
