The Bus Stops here

Phil Wainewright has some excellent points here. The premise of an ESB is the real question isn't it?

If ESB is a package, and that package is designed to enable SOA and merely to be more standards compliant and less costly than traditional EAI then yes the emergence of WS-Relability and WS-ReliableMessaging will reduce the need for a packaged ESB.

However if you consider the ESB to be a SOA pattern, and ESB vendors offering an implementaiton of that pattern, then the ESB is just fullfilling its own promise. The promise behind a standards based pattern is that certain implementation aspects will be driven to be commodity, and vendors will need to innovate to differentiate their products as they recognise each layer of the ESB should be swapable for the best of breed providers solution.

Further, custom adapters will not go away anytime soon, there are a multitude of proprietary legacy systems that adapters are the only connectivity and EAI vendors are the niche players.

Additionally if you consider ESB as a pattern, then it is very acceptable to have a few implementations of that pattern and federate accross them (its part of the pattern). The standards provide the glue to integrate and federate an set of ESB's.

Continued standards development is key for driving the ever increasing value of SOA-enabled enterprise, and those enterprises at the bleeding edge are realizing the ROI to justify some level of errosion of competitive advantage in trade off for standards that enable further competitive advantage development.

So its not the end of the road for the service bus, the bus just got a new paint job.