Almost every host advertises 99.9% uptime. It is close to meaningless as published, because the interesting parts are all in the definitions.
What The Number Leaves Out
Measured over what period? A month resets the maths every thirty days. Measured from where. Inside their own network, or from outside where your customers are? Measured against what, exactly. is the server responding to a ping enough, or does the site have to actually load?
And what counts as excluded? Planned maintenance usually does not count. Neither does anything attributed to an upstream provider.
With generous enough definitions, almost any service is 99.9%.
Why A List Is Harder To Fake
An incident history has dates, durations and descriptions. It can be compared against what you experienced. If your site was down on a Tuesday and the history does not mention Tuesday, that is visible.
The awkward part, and the reason most providers do not publish one, is that it also records the incidents that reflect badly. The ones caused by our own change, our own missed monitoring, our own slow response.
Those are exactly the ones worth publishing. A provider who only records upstream failures is writing marketing.
What We Publish
Live status per service, checked from outside our network. Real uptime figures rather than an aspiration. And the incident list, with what happened, how long it lasted, and what we changed afterwards.
Including the ones that were our fault. Particularly those, because a fault we have written down publicly is one we are considerably less likely to repeat.
How To Use It
Look at any provider's incident history before you sign up. If there isn't one, ask why. A completely empty history for a service running several years is not a sign of perfection, it is a sign that nothing is being recorded.
