The ESB is dead. Long live the ESB.
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
While major vendors like IBM and BEA Systems Inc. have rushed into the enterprise service bus market this year, more established ESB vendors have been rushing to recast their products as far more than a method of middleware connectivity. In fact, the latest announcement coming out of Fiorano Software Inc. brands its overall product as a service-oriented architecture platform, not an ESB.
The difference, according to Fiorano CEO Atul Saini, lies in business process orchestration and prebuilt Java Connector Architecture (JCA) components. Instead of supplying pure connectivity, the Fiorano SOA Platform 2006 looks to provide some intelligence around that connectivity and even hand over some basic business components that customers can reuse when creating Web services.
"The ESB by itself is of limited value," Saini said. "It adds intelligent direction routing to what we've seen from previous generations of middleware and that's it."
But Saini isn't throwing the ESB under the bus. Fiorano's ESB lies at the heart of its SOA platform.
"Messaging has to be done if you're using different machines," he said. "Once you start using multiple endpoints something has to be in the middle."
The ESB provides the nervous system for the visual mapping tools, the JCA components and the Business Process Execution Language (BPEL) service modeling that form Fiorano's SOA platform. Like competitor Cape Clear Software Inc., which will announce its latest set of upgrades later this month, Fiorano plans to attack the market with the message that all of the associated functionality it has built around the ESB will make SOA adoption a simpler, less expensive proposition.
ZapThink LLC analyst Ron Schmelzer said the ESB market never attracted a lot of money in the first place and now that IBM, BEA and open source ESBs are entering the space, it leaves even fewer pickings for the companies that built the category. He added that ESBs seem to have lost some cache as IT shops grapple with the larger issues of building to a service-oriented design.
"The ESB term has turned out to be a real quagmire," Schmelzer said. "It's somewhat, but not really related to SOA. The real challenge of SOA is not that you need more messaging. Where customers really need help is in managing all the meta data around the services they're creating."
Certainly that's where the BPEL server comes in for Fiorano, but Dennis Callaghan, an analyst with The 451 Group, argued that's not enough to qualify as a full SOA platform.
"Fiorano has added a better BPEL engine to their enterprise service bus and now calls it an SOA platform," he said. "Does that mean they now have a complete platform for managing an enterprise's service-oriented architecture? Not really. I would argue that you still need Web services management, a registry and security to make that claim."
Callaghan added that no one vendor has achieved all that at the current moment. On top of business process management, he believes business activity monitoring, Web services design and SOA governance will be areas into which ESBs will spread as the basic category gets commoditized.
Meanwhile, Schmelzer argued that the change needs to happen rapidly.
"Vendors have to prove that they're relevant to service-oriented architecture," he said. "It can't just be the same old stuff you've been selling for years in a repackaged form. What's different about your product in the last six revolutions that's different than it was for Web tier?"
For ESBs that case for relevance seemingly will be made around how quickly they can "snap together" new services, as Saini put it, and incorporate them into an SOA. The future of the ESB looks like it will be as part of a package that makes life easier for customers.
"Customers want to build service-oriented architectures without having to pay for consultants at every step," Saini said. "The dependency on consulting can only be reduced if you reduce the bloody complexity. Without that, you won't save time or money."