Baselining Performance Levels

Performance test results, gained by testing an application's performance prior to its live debut, can provide appropriate baselines unique to a specific application environment. For example, if test results show that a typical purchase transaction consistently requires 45 seconds to process, then 45 seconds would become the logical baseline to measure against in the live environment. After establishing acceptable levels of application performance, organizations possess key information from which to draw comparisons.

Baselining is the key to ascertaining a network's load and performance capabilities going forward; this is known as capacity planning. It enables network administrators to understand their current level of capacity and performance, a requirement for understanding the impact of new network application deployment or additions to deployed applications. Baselining should be done on at least a quarterly basis to identify capacity and trends.

Performance baseline objectives can be further normalized within the live environment to more accurately represent response times based on times of day and/or geographic location.

At the start of business, for example, most organizations experience a peak of network traffic as users begin the day by logging on to systems. This increase in load likely represents a peak for the day, with general usage recorded at a much lower load. It is unwise to use the peak data in isolation without considering that the peak is a once-a-day occurrence. Therefore, the baseline should show a peak at the start of the business day diminishing to a lower level throughout the day.

It may also be necessary to take into account different geographic locations and work practices. For example, start of business on the East Coast is 3 hours ahead of start of business on the West Coast, so the load experienced because of users logging on to a centralized system will see different peaks with regard to users on the East Coast and West Coast.