End of the road for ESB

Enterprise service bus, the buzzword that's enjoying the peak of its hype cycle just now, is about to become obsolete. Once it starts to fall out of favor, I'm betting that it'll be downhill all the way. I doubt many people will be talking about ESB in a year or two.

This week saw the launch of the first SOA infrastructure product to offer robust, enterprise-class messaging built solely on web services standards. Blue Titan's Network Director RM, which is due to ship early next year, embeds queue technology in its web services routers to guarantee delivery of messages, and is also fully compliant with the latest draft of WS-ReliableMessaging.

At a stroke, the notion of ESB becomes redundant. Why double up messaging capability if you have it integrated into the SOA infrastructure already? "We believe the market for ESB is transitory," Blue Titan's director of product marketing Chris Schin told me last week. I wholeheartedly agree with him.

At Gartner's Application Integration and Web Services Summit this week, the research organization's chief cheerleader for ESB, Ray Schulte, admitted the critical flaw in the ESB concept: "The newest integration headache might be over integrating proprietary ESBs," he said, according to SearchWebServices. What a misnomer it is to call this technology an enterprise bus when it seems that most enterprises will have several of them — if not dozens — each addressing a separate archipelago of integration.

Blue Titan readily admits that its product doesn't replace all of the functionality that's typically included in a packaged ESB. But then Blue Titan's aim is to give its customers an SOA fabric that straddles the entire enterprise, not sell them an extra piece of integration middleware that will still require further mediation.

An ESB looks attractive when compared to costlier, more proprietary EAI solutions. But unless it's based purely on web services standards, then it still involves integration compromises that customers may later come to rue. Some will find they have to double up the (usually JMS-based) messaging of the ESB with subsequent web services-compliant messaging technology; most will find themselves 'tripling up' as they layer web services above both JMS and whatever proprietary messaging they already had installed previously. Now that WS-ReliableMessaging is production-ready (along with its Sun-backed rival, WS-Reliability, which OASIS ratified this week), we're close to being able to build a reliable, secure services infrastructure that's entirely based on web services standards, and customers can choose to skip the ESB step.

SearchWebServices chose the right headline: The bus stops here. Indeed. My prediction is that many customers will be hopping off the packaged ESB bandwagon, rather than clambering on.

posted by Phil Wainewright 1:21 PM (GMT) | comments | link