What a network SLA covers
A service level agreement sets measurable commitments for a network service. For enterprise WAN, DIA, MPLS, Ethernet, SD-WAN or managed Wi-Fi, it typically covers availability, repair time, latency, jitter, packet loss and support response.
The SLA is both technical and commercial. It defines performance targets, measurement methods, escalation paths and remedies such as service credits. It rarely compensates the full business loss from downtime.
The metrics that matter
| Metric | Meaning | Why it matters |
|---|---|---|
| Availability | Percentage of time the service is usable. | Good for outage tracking, weak for brownouts. |
| Latency | Delay between source and destination, often round-trip. | Affects voice, video, VDI, trading and SaaS responsiveness. |
| Jitter | Variation in packet delay. | Critical for voice and video quality. |
| Packet loss | Packets that do not arrive. | Causes retransmits, poor calls and unstable sessions. |
| MTTR | Mean time to repair or restore. | Shows operational recovery expectation. |
| Time to respond | How quickly support acknowledges and acts. | Important when the provider disputes whether the SLA is breached. |
Measurement details decide the value
Always ask how the metric is measured. Is latency measured one-way or round-trip? Is packet loss measured between customer edge devices, provider POPs or internet destinations? Is availability measured per circuit, per site, per month or across the whole contract?
Measurement windows also matter. A monthly average can hide repeated short brownouts. Real-time applications often need percentile reporting, such as p95 latency or jitter, plus clear thresholds for packet loss bursts.
Service credits and exclusions
Most SLAs offer service credits, not cash damages. A one-day outage may produce a small percentage credit on the monthly recurring charge, even if the business impact is much larger. Credits are useful leverage, but they are not business continuity.
Exclusions commonly include customer equipment, planned maintenance, force majeure, third-party faults, power failure at the customer site, inside wiring, denial-of-service attacks or cases where the customer did not open a ticket promptly. Read these carefully.
Different services need different SLAs
- DIA. Focus on availability, repair time, latency to provider backbone and packet loss.
- MPLS / Ethernet WAN. Add class-of-service commitments for voice, video and business-critical apps.
- SD-WAN. Distinguish provider underlay SLAs from overlay application experience.
- Managed Wi-Fi. Include AP uptime, controller availability, support response and user experience metrics.
- International connectivity. Ask for route, cable diversity and expected latency by path.
Network SLA buyer checklist
Sources and further reading
- ITU-T Y.1541 network performance objectives for IP-based services
- Cisco IP SLAs Configuration Guide
- Cisco IP SLAs overview
- TechDirectory: SD-WAN enterprise guide
- TechDirectory: LAN vs WAN basics