10.5.2 Scenario overview
The goal of our scenario is not to demonstrate performance tuning Trade 6.1, but to demonstrate the effect of dynamic resource allocation. Therefore, we did not fine-tune Trade 6.1 for performance. We installed and configured Trade 6.1 "out of the box" without any specific tuning. As a result, the performance figures shown in this chapter are not realistic and should not be used to reflect the performance of the application or hardware.
In a production environment, your deployed application may need to support thousands of concurrent users. This is normally supported by distributing the load across a number of Web servers and clustered appservers. For our scenario, however, we did not cluster our Trade 6.1 and did not use a Web server. We used the internal Web server for server1 at the default port 9080.
The key feature we demonstrated is the use of DLPAR to dynamically increase CPU resource by adding more processors to the running LPAR. We deliberately assigned our LPAR with a minimal number of processors and ran a load test against it. The CPU utilization and response time of Trade 6.1 were recorded. While the load test was still running, we dynamically allocated more processors to the LPAR to demonstrate the improved CPU utilization and subsequently the improved response time.
To simulate a high workload to be applied to Trade 6.1, we needed a load testing tool to measure the throughput and response time as though it was in a real test scenario. There are a number of tools available for this purpose. Some are available without charge, and others are not. We used IBM Rational Performance Tester for creating our test case, as explained in the following section.
In the next section, we demonstrate how to create a test case for load testing Trade 6.1.
Next