Drafting SLAs: Writing Measurable Commitments

Why SLAs Must Be Measurable

A Service Level Agreement (SLA) is a contract section that defines exactly what a vendor must deliver, how performance is measured, and what happens when the vendor falls short. In data center operations, vague language such as "the vendor will respond promptly" creates disputes, delays repairs, and leaves the operator with no legal remedy. A well-drafted SLA protects the facility, the clients it serves, and the continuity of critical IT systems.

The standard framework for writing strong performance commitments is the SMART criteria: Specific, Measurable, Achievable, Relevant, and Time-bound. Every commitment in an SLA must satisfy all five criteria before it can be considered enforceable.

The SMART Framework Applied to SLAs

Specific means the clause names the exact service, the exact metric, and the parties responsible. "Network carrier uptime" is too broad; "99.99% monthly availability of the dedicated 10 Gbps fiber circuit between Meet-Me Room A and the carrier POP" is specific.

Measurable means the clause states a numeric target and explains how that number is collected and verified. Availability is typically measured as the percentage of total minutes in the reporting period during which the service was operational. Always specify whether measurement is continuous or sampled, who holds the monitoring data, and how disputes are resolved.

Achievable means the target is realistic given equipment capabilities and industry norms. Promising 100% uptime for any mechanical system is not achievable; 99.9% or 99.99% may be, depending on design. Setting unachievable targets benefits no one because vendors either refuse to sign or price the risk into the contract.

Relevant means the metric directly reflects the impact on data center operations. For a CRAC (computer room air conditioning) unit service response, measuring technician arrival time is relevant because prolonged equipment downtime raises intake temperatures and risks hardware failure.

Time-bound means every commitment specifies the reporting period (monthly, quarterly) and any time windows within which action must occur (e.g., four-hour on-site response).

Key Metrics You Must Define

Before writing a clause, choose the correct metric and define it in the SLA itself or in an attached definitions annex.

• Availability is the ratio of uptime to total scheduled time, expressed as a percentage. A 99.9% monthly availability allows roughly 43.8 minutes of downtime per month.
• MTTR (Mean Time to Repair) is the average time from failure notification to restored service. It is calculated by dividing total repair time over a period by the number of incidents.
• MTBF (Mean Time Between Failures) is the average operating time between one failure and the next. High MTBF indicates reliable equipment.
• RTO (Recovery Time Objective) is the maximum acceptable time to restore a service after an outage. This is a target, not a guarantee, unless written into the SLA as a commitment.
• RPO (Recovery Point Objective) is the maximum acceptable age of data that must be recovered after a failure. RPO governs backup frequency requirements.

Structure of an SLA Section

Each SLA section should contain four elements:

1. Service description — what is being delivered and by whom.

2. Performance commitment — the numeric target, the metric definition, and the measurement method.

3. Reporting obligation — how often the vendor reports performance data and in what format.

4. Remedy clause — what the vendor owes the operator if the commitment is missed (typically a service credit calculated as a percentage of the monthly fee).

Writing Enforceable Remedy Clauses

A remedy clause must specify the trigger (which metric missed, by how much), the calculation method for the credit or penalty, the cap on total credits, and the process for claiming the remedy. Example of a weak clause: "Vendor will compensate customer for outages." Example of a strong clause: "For each calendar month in which availability falls below 99.9%, vendor will credit customer 5% of the monthly recurring charge for each 0.1 percentage point below the threshold, up to a maximum of 30% of the monthly recurring charge. Customer must submit a credit request within 30 days of the affected month's close."

Safety Considerations

When drafting SLA sections that involve physical maintenance—such as UPS battery replacement or CRAC unit service—include language requiring the vendor to follow applicable lockout/tagout (LOTO) procedures in accordance with OSHA standards before working on electrical or mechanical equipment. SLAs for electrical maintenance vendors should require technicians to wear appropriate arc-flash PPE as determined by an arc-flash hazard analysis. Requiring vendors to provide proof of safety training and compliance documentation before beginning work protects both workers and the facility.

Drafting Process Summary

1. Identify the service category and the metric that best represents acceptable performance.

2. Set a numeric target supported by equipment specs or industry benchmarks.

3. Define how the metric is measured and who owns the data.

4. State the reporting period and reporting format.

5. Write the remedy: trigger, calculation, cap, claim deadline.

6. Review every clause against the SMART criteria before finalizing.