Definition
Availability is the measure of how reliably a service is up and usable over time. It's the broad concept that uptime percentage quantifies: a service that is available 99.95% of the time was usable for all but about 22 minutes of a 30-day month.
High availability means a service stays usable even when individual components fail, typically achieved through redundancy, failover, and good operational practices. Availability is always measured against a period and a definition of "usable."
Why It Matters
Availability is what customers actually experience. They don't see your architecture — they see whether your service works when they try to use it. It underpins SLAs, shapes reputation, and for revenue-critical services directly affects income. Measuring it keeps reliability honest.
How It Works
Availability is calculated the same way as uptime: (time available ÷ total time) × 100. Monitoring tools sample the service at intervals and aggregate successes and failures. "Available" should mean a correct response, not merely a reachable host, so the metric reflects real usability.
Real-World Example
A SaaS dashboard is checked every minute for 30 days. It's unreachable for 13 minutes and serving errors for 9 minutes — 22 minutes unavailable. Availability = (43,200 − 22) ÷ 43,200 × 100 = 99.95%.
Best Practices
- Define "available" as a correct response, not just a reachable server
- Measure availability per critical user flow, not only the homepage
- Pair availability with response time to capture degraded-but-up states
- Report availability with its period so the number is meaningful
- Design for redundancy on the components that affect availability most
Common Mistakes
- Equating a reachable host with an available service
- Reporting availability without the measurement period
- Ignoring partial outages that affect only some users or regions
- Measuring availability too infrequently to be accurate
- Confusing internal (provider-reported) availability with what users experience
In Monitoristic
Monitoristic shows your availability as an uptime percentage with a 30-day history bar per monitor. Because each check verifies the expected status code, the number reflects genuine usability rather than mere reachability.