Strengthen service-level agreements

Buying network infrastructure is hard enough, but making sure your service provider lives up to its commitments is an ongoing challenge that takes time and money.

Getting what you paid for goes well beyond simply verifying a service provider's position in the marketplace and the quality of its equipment and services. It requires negotiating prowess, vigilant measuring and constant enforcement. Everything hinges on crafting a strong service-level agreement (SLA).

The first step begins with finding out what levels of service your organization actually needs to do its business, not what service providers offer in their standard agreements. Knowing what your infrastructure has to be able to support will (or at least should) define what your SLA looks like.

Advertisement:

If a guaranteed level of service is important, be sure that the service you buy can reasonably be guaranteed. Asking a service provider to artificially guarantee a service that it cannot support doesn't benefit you - it actually costs you more money.

A typical network services SLA probably covers metrics such as availability, latency and throughput. It may also include specifications for mean time to respond, mean time to repair and problem notification/escalation guarantees.

The more comprehensive SLAs include requirements for installation lead times that range anywhere from 30 to 90 days based on geography; service changes ranging from five to 90 days depending on type of change; billing dispute-resolution timing requirements; and account management responsibilities such as training, documentation, and regular technology and product briefings over the life of the contract.

Still other SLAs are more than 50 pages long and protect everything under the sun.

So why not invest the time upfront to negotiate the biggest, baddest SLA on the block? Because the more comprehensive the SLA, the higher the cost of the service. The real question to consider is, "Does it really make the service better?"

The truth is that SLAs are nothing more than insurance policies. Just as life insurance doesn't guarantee life, SLAs don't guarantee levels of service. They provide you with compensation in case something goes wrong.

Carriers don't redesign their equipment to a higher standard simply because a customer demands that it perform at a certain level, nor do they rearchitect their networks because a handful of customers require a 99.999% availability guarantee. Instead, vendors determine the cost of not meeting the SLA and factor that into the overall price.

The purpose of an SLA is to protect your company against the worst case. Effective SLAs do more than get a nominal credit back - usually 5% to 10% of the cost of the service in the event your infrastructure fails. When written properly, SLAs give you a way to mitigate the effect of problems that harm your network.

Not all SLAs are created equally. Make sure yours buys what you're already paying for.

SLA strategies

The benefits of an effective SLA are well worth the time and effort it takes to develop one. Here are the steps to take to achieve your goals:

1. Identify service levels that your infrastructure needs so the SLA is comprehensive.

2. Design the SLA so that it clearly defines the service provider's responsibilities.

3. Negotiate the SLA with the service provider, paying particular attention to what services are being guaranteed, how they will be measured, the process for realizing agreed-upon remedies, and the amount of time the service provider has to correct problems.

4. Implement SLA measurement and enforcement tools and processes to ensure that every SLA can be measured and enforced as soon as the service under consideration is installed.

5. Enforce SLA compliance, and identify and resolve problems that arise.

what's the difference? Take a look at some of the distinctions between a carrier’s standard service-level agreement and one that better protects you. Standard SLA: Effective SLA: • Guarantee a certain level of circuit performance over the carrier’s core backbone that is consistently exceeded by its network (including availability, latency, throughput). • Guarantee a certain level of end- to-end service performance for applications. • Compensation for prolonged circuit outages is a nominal credit — usually 5% to 10% of the cost of the service. • Remedy for prolonged circuit outages is reimbursement of the cost of the back-up services used. • The protection from repeated failures is a nominal credit capped at 50% of the total monthly recurring cost. • Protection from repeated failures is the opportunity to augment/replace defective services with more reliable services from another vendor or use a different technology at no additional cost. • The options for getting out of a contract for failure over a prolonged period of time are nil. • Option for getting out of a contract for failure over a prolonged period of time is termination of the agree- ment without penalty.

Stanley is director of Greenwich Technology Partners' Network Economics practice. She can be reached at sstanley@greenwichtech.com.

Crafting service-level agreements that work
There's good news and bad news on the topic of service-level agreements.
Network World, 03/25/02.

Real-time SLA monitoring tools
A combination of continuous traffic analysis and proactive network testing is the most powerful way to prove service performance.
Network World, 01/15/01.