Cape Clear's Annrai O'Toole makes the observation that the most important letter in the ESB acronym is the "S". He explains that "customers are buying ESBs because they are interested in rolling out Service-Oriented Architectures. They are not rolling out 'Enterprise-Oriented Architectures' or 'Bus-Oriented Architectures.'"
He observes that nobody is interested in buying a plain old "Enterprise Bus" (EB) that's because an EB is just a messaging product, and they have already bought at least one (or in many cases two) of them. "Messaging is yesterday's news."
A reader of this blogsite, who goes by the moniker Jorwell, makes a very compelling point along these lines: why mess around with enterprise service buses when we should set our sites on an "Enterprise Service Ferrari?"
In other words, why not direct efforts and resources to newer, simpler and elegant applications that do the job, rather than attempting to build layer upon layer of code into an already messy situation? The reader observes that "SOA seems very much like adding a whole raft of new smaller applications on top of the large applications and therefore creating just as much inflexibility and disorder as it is intended to fix. In some respects the names chosen seem a dead giveaway to the inherent 'geekiness' and lack of business orientation of SOA. Enterprise Service Bus? How many senior executives understand a bus as anything other than vehicle for public transport?"
Clean, efficient, elegant — with any messiness invisible to the user. That's what could turn an ESB into an ESF. "A service has a range of attributes to it, including the quality of service offered to the users of that service," O'Toole opines. "If that quality of service requires that it be hosted on top of guaranteed delivery protocol, then it is up to the ESB to make sure that the appropriate messaging protocol is selected to ensure that quality of service. However, the specifics of the messaging protocol are irrelevant to the service user. If you do want to make those specifics available to the user, then you breach every single law of abstraction known and you defeat the purpose of using an ESB in the first place."