At first, there's the mystique that it will solve every problem known to IT. Then you get to know it and learn its limitations. Once it's determined to be difficult to implement, or to trace down to the bottom line, it's given up on. Like many technological efforts that have come and gone over the years (see CORBA, e-Commerce, B-to-B), this is the fate of SOA -- at least according to pundits such as those at Burton Group, who announced the death of SOA back in January.

But John Michelsen of ITKO Software sees SOA as "the Marconi of software." Marconi, of course, said you can take a signal, put it in the air and receive it somewhere else. He was derided as a lunatic, but his ideas led to a little thing I like to call "radio." RIAs, SaaS and anything that moves away from point-to-point EDI is fundamentally a service-oriented architecture, Michelsen noted. He went on to say that banks, at a time when their stock prices are at all-time lows, are signing big contracts to implement SOA.

"If you're Wells Fargo and Wachova, you need to merge quickly and integrate in months. That's a pressure-cooker to make of lot of things work so customers see a single integrated experience. You used to have three years in IT after a merger like that; now you have months. And I'd be stunned to see an enterprise architect today say he'll build one monolithic n-tiered fat-client application with a direction connection back to [a database]."