Your pedestal is showing

An interesting discussion on why Enterprise Architecture sucks. This is particularly relevant to me--currently supporting an engineering organization of 200-400 users (depending on who you count how). We're big enough that there are forces pulling towards "enterprisey" solutions.

I'm particularly fond of the model that says the solution is likely to be a loosely coupled federation of lighter weight applications--rather than a single monolithic design. I've been involved in too many projects that were large, lumbering, late, and short on delivery. There's no excuse for this in today's climate of open source platforms, languages, applications, and even implementations of enterprise class frameworks like like CAS, LDAP, OSMQ, etc.

One aspect of this argument that's missing is _evolution_. Delivering the 80% solution is probably optimistic. You're more likely to deliver something much short of that--say 40% and then need to _evolve_ into the 80% solution over time. Towards that end, again, the loosely coupled federation pays off as you can swap out entire sub components if necessary.