Many customers ask me “Where should I start?” after operationalization has been completed. This includes - custom dashboards, specific alarm matrices, and deployment of VirtualWisdom are complete. From a Principal Architect position, this question will crop up, many times. As we guide our customers to a balanced, hopefully, utopian destination, I typically guide our customers down this path - Health, Utilization, and Performance.
This should typically be carried out in sequence. For example, encoding errors in the FC realm - is a Health metric. You cannot measure performance, without removing all encoding errors in your environment. Measuring latency as an example, on a link that has encoding errors, will give erroneous results. You cannot measure latency, a performance metric, on a link with encoding errors.
With Utilization, specifically high Utilization, bottlenecks can occur, incurring high latency. In this case, find a way to alleviate the Utilization first within the application/host path before working towards solving performance issues.
Once the above has been remediated, then the next step is to define an SLA. This is very dependant on the application and more importantly, the application-defined thresholds. This will be the basis for monitoring and alarm thresholds. VirtualWisdom can time each exchange, at the sub-millisecond level. This way you can prove that you are within SLA limits.
To all, I pose this question; What process have you taken, to achieve optimal Health, Utilization, and Performance?