Introduction to the WebSphere Application Server performance tests

Edit online

Objectives

The objectives1 of this project were to gather Linux® end-to-end measurements for a typical transactional workload on WebSphere® Application Server using the Trade application. We gathered performance data from all of the components in the transaction path, from the user accessing the WebSphere® Application Server system to the database. The products in the transaction path were: RedHat Enterprise Linux® (RHEL), WebSphere® Application Server, and IBM® DB2® Universal Database™ (DB2®) on Linux® for IBM System z™ . We wanted to show how this set of products, which is needed to run the WebSphere Application Server Trade 6 benchmark on Linux for System z™ , performs and how to optimize these products in a customer-like environment. We performed release-to-release comparisons, CPU scaling runs, to determine the size of the workload that can be run on a certain amount of CPUs. We also determined how the workload size can grow with an increasing number of CPUs.

This is a continuation of a previous project described at https://developer.ibm.com/technologies/linux/#wasbp

Executive summary

We did not see any major performance issues when using the new versions of WebSphere Application Server (6.1.0.11) or RHEL (5.0). We recommend using RHEL 4.5 or higher for WebSphere Application Server workloads. We were not able to obtain RHEL 4.4 data because of performance issues, which are fixed in RHEL 4.5.

For Trade application tuning:
  • We showed in a previous paper that the network bandwidth has a major impact, especially the connection between WebSphere Application Server and the database. Therefore, we used for this connection HiperSockets™ and increased the number of I/O buffers for all network connection.
  • Our middle class disk I/O subsystem with an ESCON® connected ESS F20 storage server was sufficient to provide the needed disk I/O bandwidth.
  • The CPU scalability of this workload was very good. This shows that customers with similar workloads should be able to easily react to growing business by just adding more CPUs instead of installing multiple server environments.
1 This paper is intended to provide information regarding performance of environments using WebSphere® Application Server 6.1. It discusses findings based on configurations that were created and tested under laboratory conditions. These findings may not be realized in all customer environments, and implementation in such environments may require additional steps, configurations, and performance analysis. The information herein is provided “AS IS“ with no warranties, express or implied. This information does not constitute a specification or form part of the warranty for any IBM® products.