Via Service Oriented Enterprise, I recently picked up an Infoworld blog post by SOA journeyman David Linthicum, where he makes a couple of very strange points about SOA and ESBs. It may be, of course, that the post is pure link bait: certainly, David appears to have said some relatively sane things in the past, so that might be it. If it is link bait, I'm going to fall for it now.

In his piece David quotes a representative from iTKO, who repeats the now well-understood insight that in some large organisations, there's a challenge that arises because different groups with SOA initiatives find themselves having to integrate different ESBs in order to pull together implementations for cross-silo scenarios. He then goes on to comment (I?ve taken the liberty of extracting elements of his piece below):

First, if there is indeed 'enterprise architecture' and an 'enterprise architect' then the different divisions should not be using different ESBs, or even an ESB for that matter... Second, considering that my first point is correct (which it is), why the heck are you attempting to integrate these integration engines when they should perhaps be displaced altogether. Because, call me crazy again, that would be cheaper... just to be clear, you determine your requirements, and then the solutions patterns, and then the technology.

The most charitable explanation I can come up with for David's position is that he believes passionately in the transformational power of enterprise architecture. The problem with this perspective is that even where EA is highly effective (and in many places it isn't), it can at best only be one of many forces shaping the way that IT evolves to support changing business conditions and requirements, and each force has its own vector. Some forces, like a good EA team, try and combine business and technology focus, and promote the value of global optimisations, good practice and standardisation. But most of the most powerful forces are business forces, and in 99.99% of organisations, their power, when something really big and important happens, will trump any righteous splutterings emanating from IT departments.

As any experienced IT staffer who's been on the sharp end of a big business merger or acquisition, or even a radical change of leadership, will tell you, businesses don't act like machines that EAs can simply steer so that they tend towards technology optimisation. In fact, it's the opposite: business change forces (new competitors, new product launches, new market launches, new regulations, mergers and acquisitions, and so on) will always drive entropy, tending to push IT estates towards chaos. The best value an EA team can really provide in this environment is to help the IT organisation to absorb these changes with as little stress as possible, and drive consistent, planned responses.

Still, there's always been a vocal (and quite often technically brilliant) part of the IT industry that seems to misunderstand this, and persists in maintaining that once people see beautiful models, they'll modify their behaviour to enable the models to be made real. They see businesses as deterministic machines that can be algorithmically decomposed, analysed, predicted, and shaped. I can only think that David is one of these people.

Unfortunately, no matter how clever your EA team might be, no amount of architecture astronautics can make businesses change their nature.