6.6.3 Availability measures

What this heading covers

Availability is a ratio with a political denominator. Planned maintenance, partial degradation and dependency outages can each be counted in or out, and the choice changes the number. The measure should state the service window, the monitoring points and the treatment of each exclusion. A monitor that checks one endpoint measures that endpoint, not the product. Availability also depends on the dependencies the product does not control, and the measure should say whether they are in scope.

Record position: family 18 of 40 in clause 6, under 6.6 Reliability, which carries 4 families. It is one of the 40 families a programme has to map its measures to, and its elements are listed below. The heading is 6.6.3 in the public contents; the text behind it is not public.

How a measure for this family is built

Divide the time the service was usable by the time in the service window, then state the exclusions. For a finer measure, weight by the users or transactions affected, because a failure at three in the morning is not the same as one at noon. Report the monitor and the check interval, since a five-minute check cannot see a one-minute outage.

What to put in place before publication

1. Agree the service window and the exclusion rules before reporting.
2. Record the monitoring coverage, including the endpoints not checked.

Evidence to keep

Two questions to ask

• What is the service window, and which exclusions were applied?
• What does the monitor check, and how often?

Where teams usually fall short

• An uptime percentage from a single ping, with maintenance silently excluded.

What the number cannot tell you

It cannot tell you whether the service was useful when it was up. Availability is a gate, not a quality result on its own.

What happens next

Fault tolerance is what keeps availability up when a component goes down.