What 99.9% allows
99.9% availability permits roughly 8 hours 45 minutes of downtime a year, or about 43 minutes a month. Whether that is acceptable depends on when it happens and what it affects. An SLA is only useful if it defines those details.
The clauses that matter
- What is measured. Is it the server, the network, or the service your users actually touch? An SLA that covers infrastructure but not the application can be met while your business is offline.
- How it is measured. Who monitors, from where, and at what interval? Independent, external monitoring is more credible than a provider grading itself.
- The measurement window. Monthly measurement is stricter than annual. A single four-hour outage can pass an annual target and fail a monthly one.
- Exclusions. Planned maintenance is a reasonable exclusion if it is scheduled, announced in advance and capped. Open-ended exclusions are not.
- Response and resolution times. Uptime says nothing about how fast someone picks up when things break. Look for response targets by severity, and a named escalation path.
- Remedies. Service credits should be automatic and proportionate. If you have to prove the outage yourself and file a claim within days, the credit is mostly decorative.
- Reporting. You should receive regular availability reports and a written review after any significant incident.
Questions to ask a provider
Ask for last year’s actual availability figures, an example incident report, and the longest outage they have had. How a provider answers tells you more than the headline number.
Our managed services are backed by a 99.9% uptime SLA with 24/7 monitoring. If you would like to see how we define and report it, ask us for the SLA.