← All articles
HOSTING GUIDE
Availability and SLAs: what a 99.9% promise means
An availability figure looks clear until you ask what is measured and how an outage is judged. When choosing hosting, read the SLA together with its exclusions, reporting procedure and your project's own needs.
With continuous measurement, 99.9% availability allows about 43 minutes of downtime in 30 days. That is a mathematical example, not a promise about any particular service. A provider may use another measurement period, exclude planned maintenance or measure only network and server availability rather than whether your application works.
Read the definition of an outage
Check whether the guarantee covers a database failure, an unavailable control panel, slow responses or only a complete loss of connection. Look at how an incident must be documented, how quickly it must be reported and what compensation applies. A credit on your next invoice cannot restore lost orders; assess the business impact separately.
Measure what visitors experience
Your own monitoring can regularly open a specific page and check the application's response. Important functions may warrant an additional check of sign-in or the order journey in a safe test mode. Distinguish a complete outage from slowness and keep a record of when and how widely incidents occurred. This helps you identify whether the problem lies with hosting, the application, DNS or an external service.
Monitor from several locations if visitors outside your office matter. One failed request may be a brief network error; repeated failures from two places are a stronger signal. Alerts should reach someone who can act, and monitoring should confirm when the service returns. Without those steps, you collect numbers but do not shorten the real outage.
Practical question: What length of outage is still acceptable for your project, and how quickly must service be restored? Derive your requirements from that need, not the other way around.