Beamfar

What is an SLA (service level agreement)?

A service level agreement (SLA) is a written commitment about the level of service a provider will deliver, for example availability, response time to an incident or time to restore, together with how each is measured and what happens if it is missed, such as a report, a credit or a review. The Google SRE book separates the terms: an SLA is “an explicit or implicit contract with your users that includes consequences of meeting (or missing)” the service level objectives (SLOs) it contains, most easily recognised when the consequence is financial, such as a rebate or a penalty; with no explicit consequence, it says, you are almost certainly looking at an SLO.

Definition
SLA
A written agreement stating service levels, such as availability or response time, how they are measured and what happens if they are missed.

The availability clause

An availability target is turned into allowed downtime with (1 - target) x period. For a 30-day month, a target of 99.9 percent allows 43.2 minutes. A useful clause names the measured URL and success condition, the period, the check interval and locations, what is excluded and how gaps in measurement are counted.

Agency practice

Many agencies prefer to commit to what they control: monitoring, a response time and a monthly report, and to state an uptime measure only with a definition and exclusions, because hosting and third-party services are often outside their control.

An example of wording

“Availability of https://www.example.com/ is measured every minute from two regions. A minute counts as down when both regions fail. The target is 99.9 percent per calendar month, about 43 minutes. Excluded: announced maintenance of up to two hours a month and failures of third-party services named in the agreement. Minutes without a valid measurement are reported as unknown and are not counted as up.” Every term in that paragraph can be produced from stored measurements, which is the test of a good clause.

What to check

That every number in the clause can be produced from stored measurements, and that the exclusions are written down.

Sources

Read on . Standards and vendor pages change; confirm the current text before you rely on it.

Guides that use this term

  • AvailabilityAvailability is the share of a period a service worked. The formula, how it converts to minutes of downtime, and why a chain of services is less available.
  • Uptime monitoringUptime monitoring requests a website from outside at intervals and alerts when it fails to answer. What it measures, what it misses and how it is confirmed.
  • Care planA website care plan is a recurring service in which an agency monitors, updates, backs up and reports on a client site for a monthly fee. What belongs in one.

Know first. Know why. Tell your client before they call.

Paste your client list. Both regions start checking.

No card for the trial.