As you can tell we've been all Cloud and Virtualization, and not much SOA on this blog lately. And there's a good reason. Just commented in what may be the last of the eBizQ expert forums on why SOA (which stands for Service-Oriented Architecture for those who forgot about it) may or may not have failed in the past as an architecural approach to delivering business applications.
The Forums Editor Peter Schooff asks "What reason do you give for the initial failure of SOA?" citing a recent update from Burton about SOA "rising from the ashes." Well that is one way to look at it if you also stated that "SOA is Dead" a couple years ago. (Of course, we understand both statements were not made in all seriousness and they were circulated through the pundits with healthy irony...)
Here's what I said about it:
We've gone from SOA, to SO-What. I agree that it never really failed, and it is quietly contributing real value today. It might have gotten a bad reputation just because a few early adopter companies bought a particular vendor's SOA "bill of goods" or ran down a JBOWS (Just a bunch of web services) route, and found everything didn't fit together as expected, projects ran over, etc...
That rep of a few people getting burned (perhaps by not fully embracing a SOA approach) was the only real issue with SOA. 3-4 years ago we were finding some IT leaders stating "I'm NOT doing SOA - DON'T mention it again in this room, we are doing continuous integration," but despite avoiding the term they were indeed moving down a SOA path. Now, major enterprise IT does not need to make this distinction.
I thought Michael Poulin and Tarak Modi made good brief points too. Hey, if you think SOA was a failure, chances are you weren't doing it right - you weren't actually allowing the software to be delivered in a services-oriented way, you were accepting some grand vendor lock in instead of allowing for some platform independence, you were trying to dictate an all-or-nothing strategy with partners trying to connect to you - the list of ways you can fail at SOA were quite long.
And that's before we even address the issues of maintaining continuous quality and validation in the face of constant change, something that SOA makes much more complex. Or talking about how some of those service dependencies can become huge constraints to the software lifecycle that we need to virtualize in novel new ways.
We enjoy having a soundingboard for this stuff, as we watch each new IT strategy or technology get pulled by the pundits from hope, to disillusionment, and if it works, back to hope again. There is real value being proven every day by companies who let sound architecture and focused discipline drive their SOA initiatives, even if we don't talk about it that much anymore.
(Picture: I took this in 2006 at Prague Castle, a Kafka-esque skull riding on a man figure)
Comments