Defining Service Level Management Targets

Representative metrics include Mean Time to Repair (MTTR), Availability, Response Time, Throughput and Latency. One reason why selecting metrics is so difficult is that the systems or network manager must be confident that the chosen metrics will actually demonstrate value that is obvious to end users. But does that mean that end users should be brought into the decision process?

For a service level management initiative to really succeed, heavy emphasis needs to be placed on end-user input. The risk, however, is placing too much emphasis on what end users say, because they may wind up picking the "hottest" metric, or one based on the last particular problem they had, rather than on measurements that are ultimately more useful over the long haul or for improving global network efficiency.

Our recent experience shows a potentially more effective way. Rather than an approach based solely on prioritizing service levels, we found that what really worked was a correlated combination of current service usage, service outage hours per month, current end user service satisfaction ratings and service level metric prioritization criteria.

In general, service level metric priority ratings were far higher for those services that also had greater than average monthly usage, higher than average outage intervals and lower than average service satisfaction ratings than for those that did not. By taking this approach, we ensured that service level selection was based upon actual service usage and the cost of downtime rather than the sexiest metric of the day. We expect the result to be an effective implementation of service level management in the way that it was intended to work - improved service quality in response to demonstrated need. System and network managers faced with service level management implementation should consider taking a similar approach for prioritizing which service level metrics make sense for their users.

The result will be a service level initiative that is far less based on superficial end user perception and far more effective in addressing real end user need.