HP Integrity NonStop J-Series - Manual (Page 31)

Guardian Performance Analyzer (GPA) User Guide
– (544541-006) Page 31 of 131
Establishing the Measurement Period
The purpose of this step is to establish the length of the following two time periods for
running Measure:

The measurement window: This is the total duration of a single Measure run that
collects performance data for the system.

The data collection interval: This is the length of each time period within the
measurement window for which Measure collects and reports the data. For example,
an eight-hour window and one-hour interval provides Measure data in one-hour
segments.
In establishing these periods, keep in mind the following:

The measurement window depends on how the system is used and should normally
equal the daily period of continuous use. Thus, unless you already know how the
workload on the system is distributed over a 24-hour period, you should take
measurements over the entire 24 hours. However, if you know, for example, that the
system is used fairly heavily for a certain part of the day and is idle or used very
lightly the rest of the time, you can narrow the measurement window to just the busy
period.

To keep the Measure data files from becoming too large and difficult to handle, you
should break up longer measurement periods into smaller windows. For example, if
you‟re measuring over 24 hours, break up the measurement period into three eight-
hour windows or four six-hour windows.

The data collection interval should range from 20 minutes to two hours. It should be
of such duration as to provide a suitable indication of how the system load is
distributed over the measurement window. For example, a one-hour interval should
serve well for an eight- to 12-hour window. If the measurement window is
comparatively short, a 30-minute interval should do.

If the system has periods of different use (perhaps online transaction processing
during the day and batch processing at night) you should run measurements during
each period and perform a separate GPA analysis for each period.

If system use varies substantially from day to day, you should probably take
measurements over a longer time span, for example, every day for a week or a
month.
As an example, assume that the system is in continuous use for 24 hours a day, that you
have no idea how the load varies over this period, and that the 24-hour load distribution
can vary substantially from day to day. In this case, you should take measurements
over 24 hours every day for a week or two, or even for a month if the system load varies
significantly over longer periods of time. To keep the Measure data files from becoming
too large and unwieldy, you should break up the 24-hour measurement period into three
separate Measure runs and specify an eight-hour window and one-hour interval for
each run.