Setup guide

Appendix A Avid Performance Meter
42
Performance Graph
The graph of results over time can help experienced users obtain
information about system performance.
The following are a few items to look for:
• Trends — Does the system perform at a consistent rate over time? Is
there an upward or downward trend in performance measured over
time? If you notice a trend, it might be helpful to perform testing over a
longer period of time to determine if the trend continues over longer
durations or if a recurring pattern emerges.
• Patterns — Is there a momentary drop in performance that occurs at
consistent or varied intervals? Patterns can emerge as a result of
implementation details and the configuration of a particular system
(for example, caching strategies or physical memory), or they might
indicate a load on a system that is occurring periodically.
• Changes in performance as load varies — What occurs when a
subsystem is placed under a load by other users (or Avid Performance
Meter is running on another client)? It might be easier to observe these
interactions by inspecting the plot of performance results over time.
Measurement Strategy
Throughput measurements are plotted periodically (every n seconds) and
represent a snapshot of performance at the time of the plot. This is done to
limit the amount of data that is plotted, which otherwise could include a
very large number of data points. As a result, the performance graph could
miss a momentary drop or increase or a peak in performance that occurs
between plot points. The average rate calculation reflects the impact of
such a dip or spike, even if the actual size of a momentary variation does
not appear in the graph.