Intel 170 Servers WebSphere MQ for iSeries, Performance Improvements for WebSphere MQ V5.3 CSD6

Chapter 12. WebSphere MQ for iSeries

12.1 Introduction

The WebSphere MQ for iSeries product allows application programs to communicate with each other using messages and message queuing. The applications can reside either on the same machine or on different machines or platforms that are separated by one or more networks. For example, iSeries applications can communicate with other iSeries applications through WebSphere MQ for iSeries, or they can communicate with applications on other platforms by using WebSphere MQ for iSeries and the appropriate MQ Series product(s) for the other platform (HP-UX, OS/390, etc.).

MQ Series supports all important communications protocols, and shields applications from having to deal with the mechanics of the underlying communications being used. In addition, MQ Series ensures that data is not lost due to failures in the underlying system or network infrastructure. Applications can also deliver messages in a time independent mode, which means that the sending and receiving applications are decoupled so the sender can continue processing without having to wait for acknowledgement that the message has been received.

This chapter will discuss performance testing that has been done for Version 5.3 of WebSphere MQ for iSeries and how you can access the available performance data and reports generated from these tests. A brief list of conclusions and results are provided here, although it is recommended to obtain the reports provided for a more comprehensive look at WebSphere MQ for iSeries performance.

12.2 Performance Improvements for WebSphere MQ V5.3 CSD6

WebSphere MQ V5.3 CSD6 introduces substantial performance improvements at queue manager start and during journal maintenance.

Queue Manager Start Following an Abnormal End

WebSphere MQ cold starts by customers in the field are a common occurrence after a queue manager ends abnormally because the time needed to clean up outstanding units of work is lengthy (or worse, because the restart does not complete). Note that during a normal shutdown, messages in the outstanding units of work would be cleaned up gracefully.

In tests done in our Rochester development lab, we simulated a large customer environment with 50-500 customers connected, each with an outstanding unit of work in progress, and then ended the queue manager abnormally. These tests showed that with the performance enhancement applied, a queue manager start that previously took hours to complete finished in less than three minutes. Overall, we saw 90% or greater improvement in start times in these cases.

Checkpoint Following a Journal Receiver Roll-over

Our goal in this case was to improve responsiveness and throughput with regards to persistent messaging, and reduce the amount of time WebSphere MQ is unavailable during the checkpoint taken after a journal receiver roll-over. Tests were done in the Rochester lab with several different journal receiver sizes and various numbers of journal receivers in the chain in order to assess the impact of this performance enhancement. Our results showed up to a 90% improvement depending on the size and number of journal receivers involved, with scenarios having larger amounts of journal data receiving the most benefit. This

IBM i 6.1 Performance Capabilities Reference - January/April/October 2008
© Copyright IBM Corp. 2008	Chapter 12 - MQ Series	174