Service level agreements and service level objectives
Domain 4 is the most fact-heavy: what goes into an SLA, what makes an SLO usable, and how the agreement chain fits together.
SLA versus SLO
An SLA is the documented agreement with the customer. An SLO is a single measurable target - availability, response time, resolution time, throughput. An SLA normally contains several SLOs; SLOs can also exist internally without a customer-facing SLA. The SLR is the input requirement, the OLA the internal promise that makes an SLO achievable, the UC the external one.
What a usable SLA contains
Parties and scope; services covered and service hours; the service levels themselves; responsibilities of both sides; exclusions such as agreed maintenance windows; the measurement method, data source and reporting period; reporting frequency and format; escalation and breach handling; review cycle and change control.
What makes an SLO measurable
Name the metric, threshold, population, window and source: “95% of priority-1 incidents restored within 4 clock hours, measured from ticket creation to restoration in the ITSM tool, reported monthly.” Without all five, the target is an opinion.
Worked example
“The service will be available at all times and faults fixed quickly” cannot be measured. Rewrite: “Availability of the order-entry service is 99.9% per calendar month, excluding a published Sunday 02:00–04:00 maintenance window, measured by synthetic login every minute from two locations; priority-1 incidents restored within 4 hours for 95% of cases.” Both parties now know what is measured, how, and when it is excluded.
Traps
Treating SLA and SLO as synonyms; quoting 99.99% with no measurement point; accepting planned-downtime exclusions so wide they hide outages; signing targets the delivery capability cannot support; forgetting that targets must be both achievable and affordable.