I just returned from the SOA Executive Forum in New York, where I took part in two panels. I thought both went well.
One highlight of the trip was that I finally got to meet Jon Udell, who not unexpectedly is really cool. He's every bit as bright in person as he is in writing, and is a pleasure to just hang out with and chat. He moderated the first panel, on which we spent quite a bit of time discussing the "document-oriented vs. RPC-oriented" web services issue, and also talked some about the realities of having to integrate with legacy systems with multiple protocols, multiple transports, and multiple transfer syntaxes. Bob Glushko, Scott Metzger, and Tim Ewald were the other panelists, and given that we're all experienced practitioners, overall I thought the panel gave out very pragmatic and useful messages. Scott's messages about the real-world work his company has to do to interface and interoperate with other partners in the financial space was invaluable, as were Tim's lessons learned from his work on MSDN, and Bob's focus on document-oriented services was insightful and refreshing. We also discussed some of the social issues that play a big part in building and running successful service-oriented systems.
A common theme that I heard on both panels is that many people are fed up with RPC and X/O mappings. That's good news!
The second panel, which Rohit Khare moderated, focused on standards. There was general agreement that there are too many WS-* specifications -- no surprise there -- and that their mere existence caused a lot of confusion in the user community. Listening to panelist T.N. Subramaniam was a delight, as he focused on his experiences with building real systems by taking the simplest approach (aka DTSTTCPW). It's always nice to listen to people like him who actually build working production systems. The other panelists, David Linthicum and Toufic Boubez, also added some good insights. David stayed true to his useful and effective real-world theme. Toufic, probably the most pro-standards panelist among us, related some of his successful experiences from his UDDI standardization work, but given the disdain for the way our industry develops standards that T.N., David, and I apparently share, I think Toufic was a little outnumbered. Nothing personal, Toufic! :-)
I explained to the audience that I personally hate standards work but see it as a necessary evil, and that part of the reason I dislike standards work is that many standards body participants don't actually write code or build any real systems. This drew some disagreement from the audience because some felt that I was unfairly painting all standards developers that way, which certainly wasn't my intent, given that I've met many bright and capable people through standards work. Unfortunately, though, my POV is based on real-world experience. I find it very frustrating to sit on standards groups that have members who insist on making the most noise despite the fact that they have little to no real-world experience, and thus should let others do the technical heavy lifting.
I also explained that I thought users should be much more involved in standards, so that the approaches they find useful are what actually gets standardized, rather than standards groups being vendor-driven design-by-committee efforts, as they sometimes are.
All in all, I had fun, learned a few things, and hope that others learned some things from me. Thanks to Jon, Rohit, and InfoWorld for including me.
