HP-UX Workload Manager Toolkits User's Guide

HP-UX WLM Duration Management Toolkit and HP-UX WLM Toolkit for Base SAS Software

How do I use DMTK / SASTK?

Chapter 6134

Example express lane: shares-per-metric allocation

This example (/opt/wlm/toolkits/duration/config/expressconf_shares.wlm)

creates an express lane that receives more CPU resources with each

process that is active in the workload group.

# Name:

# expressconf_shares.wlm

# Version information:

# $Revision: 1.8 $

# Caveats:

# DO NOT MODIFY this file in its /opt/wlm/examples/wlmconf location!

# Make modifications to a copy and place that copy outside the

# /opt/wlm/ directory, as items below /opt/wlm will be replaced

# or modified by future HP-UX WLM product updates.

# Purpose:

# Demonstrate an express lane that has a shares-per-metric

# goal. Express lanes allow applications to finish more quickly.

# A WLM workload group is needed to handle special queries from the

# marketing department. These special requests are very important

# and must be completed as quickly as possible without regard to the

# impact on the other services. From past experience, a base

# allocation of 5 CPU shares gives a good response for a single

# random query. Also, for each additional job added to the group, an

# additional 5 CPU shares are added to the group’s CPU allocation to

# maintain performance. This workload group only receives CPU when

# it has active processes.

# Components:

# The glance toolkit is used. See glance_app(1M) for more

# information on the glance data collectors.

# Allocation change frequency:

# Because these jobs are relatively short in nature, the default

# wlm_interval metric (60 seconds) is too long. We change the

# interval to 5 seconds to minimize the time before WLM adjusts the

# CPU allocations.