HOW TO SERIES  SLA: Getting it Right

Getting the best value from a service provider is largely dependent on how
good is the service level agreement (SLA) that an enterprise has with its
service provider. However, enterprises often either are not aware as to what
constitutes a good SLA or they end up signing an SLA that is loaded heavily in
favour of the service provider. Even if an enterprise gets an ideal SLA, the
service provider does not often adhere to its clauses. Moreover, there are
situations where an enterprise is not able to pin down whether the service
provider is at fault or not. The trickiest part is performance monitoring-often
enterprises and the service providers have conflicting views on whether the
performance parameters guaranteed by the service provider have been met or not.
Monitoring network performance parameters like uptime, latency etc. is in itself
another challenge.

Focus on business goals: Look at what business objectives you want to achieve
with your network. Your business objectives should form the fundamental basis of
the SLA. The SLA must have elements that support and take care of the business
objectives. For instance, if you are an online banking company with maximum
transactions between 9 am and 5 pm during the day, the SLA must specifically
guarantee better network performance and availability during these hours. There
would be no point analyzing performance on a 24-hour basis when you need the
network to perform better during these hours. It may require significant
commitment of an enterprise's time and resources to make an SLA that favors
them.

No traps please: Parameters like uptime guarantee must be calculated on a
day-to-day basis. Even while analyzing on a day-to-day basis, more weight should
be given to network performance during peak business hours. If the service
provider has guaranteed a 99.9 percent uptime and the network is down during
peak business, there is no point analyzing performance by taking into account
network uptime during non-business hours.

Say no to standard agreements: Service providers often come up with standard
SLAs loaded heavily in their favour and they also don't really measure up to
the business objectives of the client. SLAs should be initiated by the
enterprise and contain clear, specific terms and conditions that service
provider must meet.

Clearly define measurables: All measurables related to network performance
like uptime, availability, latency etc. should be clearly defined in terms of
what constitutes what. All standards and metrics to evaluate performance should
be defined without any ambiguity. Clear definition of performance objectives
will help avoid conflicts and finger-pointing. There should not be any room for
too much interpretation. Black is black and white is white, period. That should
be the attitude.

No vague promises: Service providers often make vague commitments, and at
times even committing services which they actually can't offer. As such, the
SLA should clearly spell out the minimum acceptable thresholds for all services.
For example, if a service provider is not providing a service in a particular
location where you want the service, the SLA should be clear about that.

Insert penalty clause, it pays: It is always wise to include penalty clause
in the SLAs. For instance, if the network is down during peak business hours,
the service provider must pay for loss of business. Also, SLA should clearly
spell, for instance, if ten customers were not able to access the network during
downtime, how much would be the average loss. The penalty can be calculated on
that basis.