Summary

The main focus of this chapter was performance analysis and tuning of a commercial workload. We provided some background information on this discipline as well as some guidelines for performing this type of task. We also briefly mentioned the effort that is taking place in the industry as far as standardized performance measurement of application servers is concerned.

The majority of this chapter was spent on an example that illustrated step-by-step the process of tuning a commercial workload. The following are some of the key points of this chapter:

• Performance tuning is a skill that is acquired through constant exposure to different problem situations.

• Performance tuning should be executed with a well-defined methodology.

• If there is no obvious suspect, make the problem scope smaller through an elimination process.

• Always choose the easier and faster way to get what you need.

• Always pay attention to the kernel level of your Linux installation. In this way, you will know if you are missing patches, for example.

• Do not expect a patch, a kernel-level parameter, or a tuning parameter to work for you in all cases.

• Always be aware of the changes and improvements being made by the open source community to the Linux operating system.

• Not all new hardware features can be helpfulfor example, hyperthreading. A lot of it still depends on the nature and usage pattern of your application and the middleware.