A service level agreement is a promise with consequences. A hosting company might commit to 99.9% availability for each calendar month and agree to credit part of the monthly fee if it falls short. The promise, the way it is measured, and the remedy together make an SLA.
The key word is agreement. An SLA is part of a contract, so what counts is its exact wording, not what the number looks like in a dashboard.
SLA, SLO, and SLI
The three terms are related and often confused. The Google SRE book, which popularised this vocabulary among engineers, separates them like this:
| Term | Stands for | What it is | Example |
|---|---|---|---|
| SLI | Service level indicator | A measurement of some aspect of the service | The share of requests that succeed |
| SLO | Service level objective | A target for that measurement | 99.9% of requests succeed each month |
| SLA | Service level agreement | A contract that attaches consequences to meeting or missing a level | Below 99.9%, the customer receives a service credit |
The SLI is what you measure. The SLO is what you aim for. The SLA is what you have promised someone else, with a price on failure. An objective that has no consequence attached is an SLO, not an SLA. Teams often set internal objectives stricter than the SLA they sign, so that a bad month does not turn directly into a penalty.
What an SLA usually defines
A useful SLA is specific. It normally states:
- The metric, most often availability, sometimes also response times or support response times.
- The target, for example 99.9% or 99.95%. See uptime for what such numbers mean in minutes.
- The measurement window. A calendar month is common. A yearly window allows one long outage to hide inside a good annual average.
- How availability is measured: from which place, at what interval, and what counts as a failure. This is where two parties most often disagree.
- Exclusions. Typical examples are announced maintenance, problems caused by the customer, and events outside the provider's control.
- The remedy. Usually a service credit, a percentage of the fee that depends on how far the service fell below the target, claimed within a stated period.
An example
A provider promises 99.9% monthly availability for a hosted website. In a 30-day month that allows about 43 minutes of downtime. In March the site is unreachable for 38 minutes in one incident and 14 minutes in another, which is 52 minutes in total.
(43,200 − 52) ÷ 43,200 × 100 = 99.88%
The target was missed. What follows depends on the contract: perhaps a 10% credit for availability between 99.0% and 99.9%. If one of the incidents was announced maintenance and the contract excludes it, the counted downtime is smaller and the target may have been met. The same events can produce a different outcome depending on the wording.
Why SLAs matter
For customers, an SLA is a basis for choosing between providers and a path to compensation. For providers, it forces a clear definition of the service and a way to measure it. For both, it removes a vague promise such as "very reliable" and replaces it with something that can be checked.
Common misunderstandings
- "99.9% means it will never fail." It means failures up to a limit are expected and tolerated. Even 99.99% allows several minutes per month.
- "A credit makes up for the loss." Credits are normally limited to part of the fee. They rarely cover the cost of an outage to the customer's own business.
- "My monitoring report is proof." The provider's measurement method usually defines compliance, and it may differ from yours. A third-party monitor can be a valuable record and a basis for a conversation, but whether it settles a dispute depends on what the contract says about evidence.
- "An SLA improves reliability." It only sets expectations and consequences. Reliability comes from engineering and operations.
- "Higher is always better." Each additional "nine" costs far more than the last. A target should match what the business needs.
SLAs and monitoring
Monitoring supplies the raw material for SLA conversations: when the site failed, for how long, and how often. UptimeWire provides monthly SLA reports on screen and CSV export on the Business and Agency plans. They show uptime, downtime, incidents, and response time for a calendar month, compared with a target you set. They are not available on the Free and Starter plans.
These reports describe what UptimeWire observed when it checked your configured websites. They do not decide whether a contractual SLA was met or broken, and UptimeWire does not guarantee compliance with any agreement you have with a customer or provider. That depends on the contract, including how it defines measurement, exclusions, and evidence.