WebSphere Application Server V8.5 Concepts, Planning, and Design Guide

Chapter 7. Infrastructure 171

If you need a more accurate estimation of your hardware requirements and you already have

your application, you can run a benchmark. Before you start the production, validate the

sizing with a performance test campaign.

For more information about benchmarks, see the following websites:

 rPerf

http://www.ibm.com/systems/power/hardware/notices/rperf.html

 SPEC

http://www.spec.org/benchmarks.html

 TPC

http://www.tpc.org/information/benchmarks.asp

7.4.2 Sizing dynamic infrastructures

In a dynamic cluster, the application placement controller needs sizing and application

placement decisions based on service policies. A service policy allows workloads to be

classified, prioritized, and intelligently routed, which makes a manual sizing for individual

resources unnecessary.

In a dynamic cluster, all resources that run in the applications are virtualized. They are then

created, started, and stopped as needed for the current workload. To estimate the resources

needed for a new application, use the techniques described in 7.4.1, “Sizing static

infrastructures” on page 170. After the application is running on the performance test or

production environment, the health controller can help you to decide whether the estimation

was correct.

The health controller can run predefined and custom corrective actions if the defined health

conditions for an application are not met. You can also define email notifications that are sent

out when health conditions warrant.

7.5 Monitoring

Because most WebSphere technology-based applications are web-based applications, 247

availability is essential. The tolerance of Internet users for unavailable sites is low. They

usually navigate to the next site if your site is not operable, meaning you lose potential

customers. Therefore, track and monitor the availability of your site so that you recognize

when things are going wrong and can react in a timely manner.

Efficient monitoring combined with a sophisticated alerting and problem handling procedure

can increase the availability of your service significantly. Therefore, you must plan for

monitoring and problem handling. Do not wait until your environment becomes unproductive.

7.5.1 Environment analysis for monitoring

Careful planning for monitoring is essential, and must start with a detailed analysis of the

environment to be monitored. Ensure that the full environment is monitored and that no

component is overlooked.

To analyze the monitoring requirements for your environment, consider the factors in the

following sections to give you an overview of what needs to be done.