The secret is to measure meaningful activities. A long, long time ago, we had an arrangement with a computer manufacturer that guaranteed a very high up-time percentage which was maintained. Unfortunately, our resonse time was very lousy. We should have measured response time (how quickly the network/computer could reply to customers’ queries). When we changed the metrics, we were able to identify short-comings, identify solutions and apply them.

The availability can be 99.999999%, but if response is slow (such as 4+ minutes to boot-up network computers) it is not acceptable. Up-time is important, but response time should also be included in the mix.

Define useful metrics, not just something that can easily be displayed, and really doesn’t show much.