SLA: What is a Service Level Agreement?

What is an SLA?

An SLA (Service Level Agreement) is a contractual document formalizing a service provider's commitments to its client. In the context of software development and maintenance, the SLA defines precise, measurable metrics that frame the quality of service delivered: system availability rate, incident response time, resolution time by severity, maintenance windows, backup frequency and technical support conditions.

For Belgian SMEs that outsource application development or maintenance, the SLA is far more than an administrative formality. It is the foundation of the trust relationship with the provider and the only enforceable document in case of disagreement about service quality. A well-drafted SLA protects the client while giving the provider a clear framework for organizing operations. At KERN-IT, the SLA is an integral part of our maintenance contracts because we believe transparency about commitments is the basis of a lasting partnership.

Why SLAs Matter

Without an SLA, service quality is left to the subjective assessment of each party. The client expects 99.9% availability while the provider considers 95% acceptable. The client expects resolution in 4 hours while the provider plans for 48 hours. These misalignments create tensions and disputes that could have been avoided with a clear SLA:

• Aligned expectations: the SLA establishes a common language between client and provider. Each party knows exactly what is guaranteed, what is a target and what is out of scope.
• Measurability: an SLA transforms vague notions (good service, responsiveness) into objective, measurable metrics. This objectivity eliminates subjective debates about service quality.
• Accountability: the provider is contractually committed to measurable results. Penalties for non-compliance (service credits, billing reductions) create a real incentive to maintain quality.
• Incident prioritization: the SLA defines severity levels (critical, major, minor) with distinct response and resolution times. This allows the provider to prioritize incidents based on their actual impact on the client's business.
• Dialogue framework: periodic SLA reviews (monthly or quarterly) offer a structured framework for discussing performance, identifying improvement areas and adjusting commitments if needed.

Software SLA Components

A comprehensive SLA for web application maintenance covers several dimensions. Availability is measured as a percentage of uptime over a given period. A 99.5% SLA allows approximately 3.65 hours of downtime per month, while a 99.9% SLA tolerates only 43 minutes. Each additional fraction of a percent implies increasing investments in redundant infrastructure and monitoring.

Response times define the maximum delay between incident reporting and first engagement by the technical team. This is not the resolution time but the reaction time. A critical incident may require a 30-minute response time while a minor bug may tolerate 24 hours.

Resolution times define the maximum delay to fix the problem. They vary by severity: a complete production outage may require resolution in 4 hours, a performance degradation in 24 hours and a cosmetic bug in 5 business days.

Maintenance windows define the time slots during which the provider can perform planned interventions (updates, migrations, optimizations) that may require temporary unavailability. These windows are generally outside peak hours.

Exclusions define what is not covered by the SLA: force majeure, issues caused by the client, unauthorized third-party interventions, feature requests (which fall under a different contract).

Concrete Example

KERN-IT manages maintenance for an online ordering platform for a Belgian food distributor. The contractual SLA provides for 99.5% availability (excluding planned maintenance), a 1-hour response time for critical incidents (platform inaccessible or inability to place orders) and 4 hours for major incidents (degraded functionality). Resolution time is 4 hours for critical and 2 business days for major incidents.

In practice, measured availability over the past 12 months is 99.87%, well above the commitment. Average response time for critical incidents is 22 minutes. These metrics are shared monthly with the client in an SLA report detailing each incident, its resolution time and corrective actions implemented. This rigorous tracking has reduced the number of critical incidents from 8 in the first year to 2 in the third, thanks to a preventive maintenance approach.

Implementation

1. Define key metrics: identify the service indicators that truly matter for your business: availability, response time, resolution time, backup frequency.
2. Set realistic levels: a 99.99% SLA costs 10 times more to guarantee than a 99.5% SLA. Define levels aligned with actual business needs and available budget.
3. Define severities: classify incident types into severity levels (critical, major, minor, cosmetic) with distinct response and resolution times for each level.
4. Plan penalties: define compensation for SLA breaches (service credits, billing reductions) to create a real incentive for meeting commitments.
5. Implement monitoring: deploy automatic availability and performance measurement tools to have objective, indisputable data.
6. Periodic review: schedule monthly or quarterly SLA reviews to analyze performance, discuss incidents and adjust commitments if needed.

Associated Technologies and Tools

• Sentry: real-time application error monitoring, detecting and measuring incidents before users even report them.
• UptimeRobot or Pingdom: availability monitoring tools that measure uptime and immediately alert on downtime.
• Prometheus and Grafana: a monitoring and visualization stack for tracking performance metrics (response times, resource usage) and creating SLA dashboards.
• Fabric: a deployment tool used by KERN-IT to execute fixes and updates quickly, reliably and traceably.

Conclusion

The SLA is the cement of the relationship between a technical provider and its client. At KERN-IT, we do not view it as a constraint but as a quality commitment that structures our daily work. Our maintenance contracts include precise SLAs, automatically measured and regularly reviewed with our clients. This transparency is at the heart of our long-term partnership approach: we commit to measurable results because we know that trust is built on facts, not promises.