What our uptime SLA actually promises
An SLA is only meaningful if you know what is measured, from where, and what happens when it is missed. Here is ours in plain terms.
The number
We commit to 99.99% availability per calendar month. That is a budget of about four minutes and twenty-two seconds of downtime in a thirty-day month. Planned maintenance is not excluded from the measurement, because excluding it would let us schedule our way to a perfect record.
How it is measured
Nine probes, one in each region, request a health endpoint every thirty seconds. A minute counts as down when a majority of probes fail it, which stops a single unhappy network path from either hiding a real outage or inventing a fake one.
What you get back
Credits are calculated automatically and applied to your next invoice. You do not have to notice the outage, open a ticket or quote a reference number.
- Below 99.99%: 10% of the monthly fee
- Below 99.9%: 25% of the monthly fee
- Below 99%: 50% of the monthly fee
- Below 95%: the full monthly fee
What is not covered
Anything we do not control: your application code, a plugin that exhausts memory, a DDoS aimed at an unprotected origin you have exposed directly, or a domain that lapses because a card expired. We will help with all of it, but it does not draw on the credit budget.
