after your users start failing. Drag the dashed line on the replay to set it there.
What if
- The status page stayed green
- minutes
- You would be first on
- Total impact
- Threshold to beat all six
- minutes
- Users failing
- Recovering
- Page says operational
- Page admits a problem
- The dark window
- You
The status page, against what was true
All six, ranked
| Rank | Incident | Dark | Total | At your threshold |
|---|
Under the hood
A status page is a press release with a clock on it. It fails the way any published spec does: optimistically.
It is written, not measured
Nothing on a status page is emitted by the failing system. A person decides the blast radius is now large enough to admit to, writes a sentence and gets it approved. That pipeline has a floor of several minutes on its best day.
It often shares fate with the outage
In December 2021 AWS posted that the incident was “affecting some of our monitoring and incident response tooling, which is delaying our ability to provide updates.” The instrument was downstream of the thing it measured. That is not rare.
You are on the clock before it moves
Your users started failing at the first timestamp, not the second. Every minute of that gap is a minute spent deciding whether it was you, and the only instrument that could have told you sits inside your own account.
- Median gap before the page moved
- Longest documented gap
- Total impact replayed here
Two clocks and one subtraction. The example follows whichever incident you are replaying.
The first clock: when customers started failing
Taken from the provider’s own retrospective, written afterwards with internal telemetry. It is the earliest honest number anyone publishes, and nobody outside could have known it on the day.
The second clock: when the status page said anything
Not when it was accurate or named a cause, just the first public admission that something was wrong. It is the friendliest reading of the provider, and the gap is still tens of minutes.
The gap
Second clock minus first. The window in which your dashboard was green, your users were not, and no outside source would have backed you up.
Your threshold against it
Your own paging delay, taken from the gap. A positive margin means you were first and the page confirmed what you knew. A negative one means the status page told you first.
Each has a published retrospective that states when impact began, the number nobody can reconstruct from outside. Incidents where that timestamp was never published are not here, however famous.
| Incident | Impact began | Page moved | Gap | Total | What broke |
|---|
Every timestamp, and where it came from. The model is five lines.
- Gap
- first public status post − first customer impact
- Margin
- gap − your paging threshold
- Dark
- impact → first post: console green, users failing
- Open
- first post → mitigation: acknowledged, unresolved
- Recovering
- mitigation → all clear: backlog draining
It does not measure whether the post was true. The gap closes the moment anything is admitted, however vague. “Investigating elevated error rates” during a total regional failure counts as the page having moved.
Nor your blast radius, nor malice. Two accounts in the same region had very different days, and every gap here has a mundane explanation; in at least two the tooling needed to update the page was inside the failure.
All times are UTC, converted where the source published local time. The provider’s wording is quoted verbatim where it was published, and marked as reported where it was paraphrased in coverage.
Impact start always comes from the retrospective, never a third-party detector. Downdetector spikes appear in the log as context because they are what an outsider saw, but they never set the clock.
Two acknowledgements are weaker than the rest: the AWS December 2021 first post and the Azure Front Door first advisory come from secondary reporting, and are marked reported. Nothing is scraped live; the page ships the data it draws.
Part of Infrabench
Free tools for the people who build data centers, on public data with every constant on the page.
Built by Adrian Mucha
A product designer for the software most designers avoid: data center tools, AI tooling and operator consoles where a wrong click costs real money.
- See the portfolioadrianmucha.us
- Open to new workBased in Los Angeles