WAN Services /

Crafting service-level agreements that work

By Johna Till Johnson
Network World, 03/25/02

There's good news and bad news on the topic of service-level agreements. The good news is that most large organizations have begun to define SLAs for all carrier service contracts - and they include items such as latency and availability that have a real impact on user satisfaction.

Even better, the carriers now are enthusiastically supporting these SLAs and even including them in their contracts without being asked.

The bad news is that in most cases there's no easy way to enforce them. Carriers have gotten used to contesting SLA-noncompliance claims, and when they do admit fault, they're more likely to write checks than to make lasting service improvements.

That doesn't help much. Telling the corner-office suits that you got a few months off next month's service charges doesn't win any brownie points when the real issue is poor service.

So what's a telecom manager to do? The answer is to craft SLAs that cause real pain - not merely the financial kind - in the service provider's organization. One of the clearest and most effective ways to do this is to craft a defined, escalation policy that ensures that any issues receive immediate attention from the right people inside the service provider's organization.

Here's how it works. First, define what constitutes a service "incident" to you and your users. Create severity tiers that characterize the impact of a service weakness. For example, outright service cessation - no dial tone, no packet transfer - is obviously tier 1.

But you also should consider latency spikes, which can affect user response-time. Sustained latency of 250 msec or more for periods of longer than 1 minute may constitute a tier-2 event.

Once you've defined the set of tiers, determine how quickly the carrier should inform you of an incident, how quickly it should respond to your notification and how quickly you can expect restored service.

After that, get the names, job descriptions and phone numbers for the individuals who will be directly responsible for repairs. Get the contact information for the individuals to whom they report. Then, define explicitly when the service provider will escalate an incident. The service provider should agree to inform you directly every time an incident has been escalated.

Now for the secret sauce: Explicitly specify that a senior member of management will address any issues that are not resolved in a timely fashion and to your complete satisfaction - at a time and date of your choosing.

In other words, if you are sufficiently unhappy, an executive vice president will fly out to your site to explain why your network doesn't work.

This works for a simple reason: Nobody wants to tell the vice president he has to exchange a Friday tee time for boardroom full of angry customers. The result is a more energetic effort to avoid the situation - exactly what you wanted in the first place.