What is SLA?

An SLA specifies the metric, the threshold, the measurement method, and the measurement window. It also defines exclusions: scheduled maintenance, force majeure, and failures attributed to the customer's own systems commonly do not count.

The defined consequences distinguish an SLA from a service target with no associated commitment. Service credits are common in commercial agreements, usually a percentage of fees refunded on a sliding scale by how far the target was missed. The exclusions and measurement method often matter more than the headline number.

An SLA converts reliability from an engineering aspiration into a commitment with defined consequences.

• Enforceable expectation: users have recourse rather than only complaints.
• Risk pricing: the provider prices the cost of missing the target into the agreement.
• Investment justification: exposure to those consequences funds reliability work.
• Procurement comparison: a common basis for evaluating vendors.

Service credits may compensate only a limited portion of a customer's actual loss and may also limit the provider's contractual exposure. A customer whose business was down for six hours may receive a small percentage of one month's fees.

Exclusions can also hollow out the commitment. An SLA excluding maintenance windows, degraded performance short of unavailability, and any failure in a dependency may guarantee relatively little in practice while displaying an impressive number.

An SLO is a measurable service-level target. An SLA adds an agreement with users and defines what happens when the included service levels are met or missed.

Teams often set an internal SLO that is stricter than an SLA target to create an operating margin before customer-facing consequences are triggered. This is a common risk-management practice rather than a defining requirement.