HP-UX Workload Manager User's Guide

Configuring WLM
Tuning the metrics and the SLOs
Chapter 5216
The shorter the interval, the more often WLM checks for new
performance data and alters the CPU allocation if workloads are not
meeting their SLOs. If there is no new data, WLM does nothing.
With longer intervals, WLM is more likely to receive new performance
data; however, a workload is also more likely to not achieve its SLO for
longer periods.
The wlm_interval tunable is optional. It is allowed only in a global tune
structure within the WLM configuration file.
NOTE The current WLM interval length is available to data collectors specified
in coll_argv statements. The length can be retrieved through the
WLM_INTERVAL environment variable.
Use the following syntax in a tune structure to specify the interval:
wlm_interval = number_of_seconds;
where
number_of_seconds
Specifies how often the WLM daemon checks for new
data from the data collectors and adjusts CPU
allocations. Valid values are from 1 to 86400 seconds (1
day).
The default is 60.
Typically, number_of_seconds should be 5 when your
WLM configuration contains a usage goal.
When generating audit data (with the wlmd -t option),
number_of_seconds should be no more than 60 to
ensure the audit data sampling rate is sufficiently
frequent to:
If you are running WLM within an Integrity VM
(guest), specify an interval length greater than 60
seconds. This helps ensure a fair allocation of CPU
resources for FSS groups.
• Minimize the amount of audit data lost if the WLM
daemon exits unexpectedly