WLMTK Overview: Using HP-UX WLM Effectively With Your Most Critical Applications
7
The remainder of this white paper discusses the benefits of each of these toolkits and how to use them.
HP-UX WLM Oracle Database Toolkit
HP-UX WLM enables you to place Oracle instances and other applications in their own WLM
workloads. With the instances and applications separated, WLM then manages the performance of
each instance and application through prioritized SLOs. For the Oracle database administrator, this
functionality provides the ability to allocate different CPU resources to Oracle database instances by
modifying the WLM configuration.
Oracle instances and applications are separated between separate Oracle database instances—not
within a single instance. That is, all Oracle processes related to a specific database instance must be
assigned to the same workload, including Oracle server processes (“shadow” processes) that run on
behalf of users accessing the database.
To take advantage of the ability of WLM’s capabilities with respect to Oracle instances, HP provides
HP-UX Workload Manager Oracle Database Toolkit (ODBTK), which facilitates transferring metrics
from Oracle database instances into WLM. This toolkit, which is included with WLM, works with
Oracle 8.0.x, Oracle 8.1.5, Oracle 8.1.6, Oracle 8.1.7, and Oracle 9.0.1. For information on
toolkit requirements, see the HP-UX Workload Manager Toolkits User’s Guide.
What metrics are available?
The following database metrics are available:
• Time elapsed while user-defined SQL code executes (this metric provides a response-time
measurement for the database)
• Application-specific value returned by executed SQL code
• Information retrieved from Oracle V$ tables using SQL (these tables provide dynamic performance
data for Oracle instances and enable the Oracle database administrator to see current performance
information)
Why use ODBTK?
The main benefit of using Oracle database metrics with WLM is that you can use these metrics to
manage the performance of your instances. For the instances, you can specify SLOs based on the
metrics.
The following sections illustrate a few reasons for using ODBTK to manage your instances. These
sections discuss various items you would set in an SLO in a WLM configuration file. ODBTK comes
with example WLM configuration files. These files, some of which are used as a basis for the
following examples, show how you can control your Oracle instances and their resources quickly and
easily.
The following examples focus on a single instance to simplify demonstrating what WLM and ODBTK
can accomplish. Of course, WLM is most useful when multiple database instances and applications
are running on a single system.
Set response time goals
In this first example, a workload named Instance1 has an SLO with priority 1, the highest possible
priority. However, other workloads might also have priority 1 SLOs. The workload’s CPU allocation
will be in the range of 20 to 80% of the system CPU resources, assuming enough CPU resources are
available to meet this SLO and all other SLOs at priority 1. WLM adjusts the CPU allocation for the
instance’s workload within this range so that the response time for the ”Select” database operation is
less than 10 seconds.