J2EE application servers as we know them today started their lives as middleware servers that connected back-end databases to front-end Web pages. IBM's WebSphere was no exception. It processed transactions, but it did not munch transactions as numerously and frequently as CICS and similar transaction processors. But Java was a broadly accepted standard, and that was important in WebSphere's success.
As time went by, and integration stood out as the major application server chore, IBM fit more and more diverse types of middleware servers under the WebSphere marketing umbrella.
We recently wrote about WebSphere Information Integrator in this light, and discussed the fact that it more or less started life as a DB2 Integrator that IBM decided to slip into the WebSphere product line. Its role in life is as a data middleware server. That's the latest in a line of WebSphere additions that first hit its stride a few years ago when IBM placed its MQ Series software in the WebSphere bucket.
MQ is message-oriented middleware that allows something Java -- at least before the Java Message Service [JMS] was invented) – did not have; that is asynchronous operation. MQ arose from the need among big banks and the like – big transaction consumers all— to assure operations by queuing up overflow transactions at peak times. Given the right circumstances, reliable messaging has clear benefits over remote procedure calls and other manner of distributed system. Today, big banks and the like are the main WebSphere MQ consumers, but IBM says it is finding traction in small- and medium-size businesses (SMBs) that need to queue up transactions, applications, and files inside or outside their organizations.
To go into the SMB space, IBM must make WebSphere MQ easier to use, and elements of a new release show effort there. A new version of WebSphere MQ is said to reduce the costs of on-going support for manual coding. It improves support for file transfers that are common in smaller organizations.
Moreover, so called WBI Server Express includes some out-of –the box adapters that integrate with existing applications, and interactive software wizards that support business-rule-based construction of messaging schemes. A new Eclipse-based workbench for MQ lets teams publish information to multiple queues at one time from a common tool.
There are three broad categories of improvement in this rev of WebSphere MQ, said IBM's Scott Cosby, Program Director, WebSphere Business Integration. Better administration and manageability, platform and deployment enhancements, and added Web services support are in store, he said.
"On the administrative side, we have moved the entire user interface over to the Eclipse framework," he said.
The added transfer capabilities Cosby discussed include "FTP tooling." In SMBs, particularly, he said FTP use is still common. Now those FTP data movements can be managed more easily as part of MQ. In effect, FTP transfers can be staged and scheduled to move over the MQ transport.
Among moves on the platform side, WebSphere Information Integrator has improved its links to the zSeries platform, where ever-growing loads require a larger MQ "footprint."
"When a transaction processor is down, these queues can build up with transaction data pretty quickly," said Cosby. Thus, WebSphere MQ will make available bigger queues, for bigger logs and bigger buffer pools.
On the Web services side, IBM's flagship messaging system now allows you to "flow SOAP over MQ," according to Cosby. Why? "You get this reliability that is just not there with Web services today," Cosby said.
"You can expose a Web services interface, say, to open a purchase order," he continued, "but the actual transaction will be processed over the existing MQ network." This would allow you to spread your messaging system out beyond your organization, in effect, without requiring your suppliers to become MQ shops.