WAN-SLA and MTTR: What Really Counts in the Contract

What is a WAN-SLA really?

A WAN-SLA is the contractual commitment to the quality of your site connectivity, not just a marketing figure on the quote. It defines in measurable terms which availability, latency, jitter, packet loss and repair time the provider guarantees, and what happens if it misses those values.

The decisive difference lies in the detail. Two providers can both promise 99.9 percent availability and still deliver completely different service, depending on how and where it is measured and how quickly incidents are resolved.

A solid SLA therefore always names three things: the committed values, the measurement method including measurement points, and the financial consequence of non-compliance. If one of these building blocks is missing, the promise is worth little when it counts.

Availability: why 99.9 is not always 99.9

Availability sounds unambiguous, but it is not. Each additional nine drastically shortens the permitted downtime, as the following table shows.

For a critical site, the difference between 99.9 and 99.99 percent is considerable: a good 8.8 hours versus just under 53 minutes of downtime per year. Equally important is the question of what counts as an outage at all. Are planned maintenance windows excluded? Does a partial fault with reduced bandwidth count? Is the average taken per site or across the entire network?

SAVECALL checks these definitions across providers and makes them comparable. For your sites we bundle the individual commitments into ONE unified SLA with ONE point of contact, instead of many contracts with different measurement logics.

The metrics that matter in operation

It is latency, jitter and packet loss, not availability, that decide the real quality of your applications. A line can be formally available and still unusable if these values are too poor.

For real-time services such as voice, established reference values apply, and a solid SLA must be measurable against them.

ITU-T Recommendation G.114 specifies a maximum one-way latency of 150 ms, while for most enterprise applications under 50 ms is considered excellent. A good WAN-SLA guarantees these values as a commitment with clear priority classes, not as a non-binding target range. Check whether the figures are given as a monthly average or as a percentile, because a monthly average can hide short but business-critical spikes.

MTTR: the underrated key figure

The MTTR, the mean time to repair, is often more important in daily operation than the advertised availability. It determines how long a site is actually offline during an incident, and that is exactly what the business feels.

Watch three points. First, is the MTTR guaranteed or only a target value? Second, when does the clock start, at ticket creation or at fault confirmation? Third, are response time and restoration time shown separately? A fast response is of little use if the actual restoration remains open.

For redundantly connected sites, the focus shifts from the single outage to the question of how quickly and automatically the failover kicks in. SAVECALL defines these points together with you and fixes them consistently across all carriers.

Check penalties and the fine print

Without a penalty, any SLA promise stays consequence-free. Service credits are the only part of the contract with real financial effect when the provider misses its values.

What matters is the amount, the trigger threshold and the procedure. Are credits granted automatically or only on request within a deadline? In most cases they cover only a fraction of the actual business loss anyway. Their real value lies in forcing the provider to deliver reliable quality.

In the fine print, it pays to look at the exceptions: force majeure, third-party shares on the last mile and the treatment of the access line. Especially when several carriers are involved, accountability gaps arise here that become expensive during an incident.