Monday, September 19, 2005

SOA Chaos 2

In my previous post SOA Chaos? I mentioned an SOA project whose first iteration (according to some commentators) "ended in chaos".

Jeff Schneider (Momentum SI) uses this project as evidence that Stupid People Shouldn't Do SOA.

I think this is unwarranted abuse - both of the individuals involved in the project, and of the project organization as a whole. According to the Information Week report, the project got something wrong and then corrected it. But in general it is not stupid to make mistakes. There are two real signs of stupidity - refusal to try anything new for fear of making mistakes, and refusal to learn from repeated mistakes.

Schneider proposes an intellectual division of labour: the clever people ("brains") should create services, and the not-so-clever people ("grunts") should use them. This division of labour implies that decomposition requires more intelligence than composition.

But although this may be true in some special cases, I don't think it is generally true. I think that composition and decomposition require different kinds of intelligence, which we might call analytical intelligence (good for solving software problems by decomposition) and synthetic intelligence (good for solving business problems by composition). Clearly there are lots of people in the IT industry, like Schneider, who value analytical intelligence more than synthetic intelligence. In my view, SOA requires both.

But the difficulty apparently faced by this project was the composition of loosely-coupled services into a coherent (integrated) business process. It seems perverse for Schneider to argue that this difficulty should be overcome by devoting more brainpower to the decomposition.

Technorati Tags:

No comments: