For me, JBI's advent will probably be a really good thing. SOA (Service Oriented Architecture) is a paradigm, not a product. I've bought into the SOA paradigm hook-line-and-sinker, but getting from vision to reality has proved difficult.

Web Services, by themselves, just aren't enough to base an Enterprise-wide SOA solution on:

• We need means to manage a large number of services.
• We need means to deal with multiple versions of services.
• We need reliable messaging.
• We need publish/subscribe in addition to request/response.
• We need management and monitoring tools.
• We need provisioning systems.

The scary thing about needing an ESB is in the commitment. ESBs have not been around that long, and what if you pick the wrong one?

JBI is welcome because it should increase commonality among ESBs. The more ESBs have in common, the less likely it is that we'll have to radically change our services and development/deployment processes if we have to migrate from one ESB to another.

IBM and BEA both abstained from the JBI vote, expressing the sentiment that JBI didn't really add all that much to the Java stack. I have to disagree... JBI may not include any groundbreaking concepts, but it greatly adds to value of the Jave stack by increasing this developers peace-of-mind.