Step 2:
Select Performance Metrics

This step addresses the question:

How is satisfactory performance measured for this product?

All performance metrics (measures of performance) are based on system behavior over time. There are three major classes of metrics:

• Latency or Response Time metrics measure the delay between initiating or requesting some action and the arrival of the result. For example, consider a on-line store. Some latency metrics might be:
  • Delay between the instant a user clicks on a "submit" button until the time all of a confirmation screen is displayed on the user's screen.
  • Delay between receiving an order for an item and updating the "number in stock" that's reported to other customers.
• Throughput or Capacity metrics measure the amount of work done per unit time. For the online store, one throughput metric might be the number of updates to the inventory database per second.
• Availability metrics measure how much of the time a system is available for normal operation. The term "availability" is used to describe the fraction of time a system is available. For example, if the inventory database is down for a hour a day, it has an availability of about 0.96. The term "reliability" is used to quantify time between failures.

Selection of appropriate metrics is a shared responsibility of the performance analyst, marketing team and development team. As with workloads, much of the analyst's task is help translate business-level objectives (such as "fast and reliable") to specific metrics.