My keynote at the Microsoft SOA and BP conference has generated a bit of press, including this excellent summary by Darryl Taft ("Gartner: How to Unify Your SOA Roll-Out") and this take on a quote from that article by Joe McKendrick ("Analyst: ‘you’ll all be doing SOA in 18 months whether you plan to or not’"). While both articles are accurate overall, I wanted to put the "18 months" comment in context. So I left the following comment on Joe’s post:
Joe, While I am bullish on the concepts that constitute SOA (modularity, distributability, discoverability, sharability, loose coupling), I am less bullish on the name "SOA" given how it has wound up meaning just about anything anyone wants it to mean.
As for the "18 month" comment, I was using "SOA" somewhat tongue-in-cheek. What I was referring to was the SEC announcement in 2008 that between now and 2011 it will be phasing in mandatory use of XBRL for financial reporting into the SEC.
Since the audience was primarily composed of IT employees of public companies regulated by the SEC, I thought I would highlight that "everyone" will have to use the automated SEC reporting service to deliver their 10Qs etc. in the form of XBRL. This is what I meant by "you’ll all be doing SOA in 18 months or so".
But of course, one isn’t really "doing SOA" unless the system one creates using services is modular, distributable, discoverable, sharable, and loosely coupled — something I pointed out later in the talk.
Feel free to add your 2 cents either here or on Joe’s post.
3 responses so far ↓
1 wolfgang roy schulte // Jan 31, 2009 at 11:46 am
This debate has more heat than light because people are using different definitions of SOA. If you asked Anne Thomas Manes, “Do you believe that applications that are modular, distributable , discoverable (have interface metadata), swappable (interface is separate from the implementation), and shareable/reusable are ‘dead’? She probably say “No, those apps are very much alive, they are best practices.” But that’s Gartner’s definition of SOA, so she actually might agree with Gartner if she was using the same definition.
She says ‘SOA is dead’ because she is thinking only about RPC-style SOA. In her book “Web Services: A Managers Guide” she speaks highly of SOA – but she implies that all SOA is RPC-style SOA. She doesn’t count REST-SOA, event-driven SOA or P2P SOA as “SOA” and her book doesn’t even mention these other styles of operation to the best of my recollection (I just re-skimmed her book and couldn’t find any mention of them, and they were not in the index either). But now, several years after she wrote the book, she has apparently lost faith in RPC-style SOA.
In reality, most of us at Gartner would agree with Anne that the other (non-RPC) styles of SOA are on the ascendancy. We’re just saying it in a more measured manner.
No one thinks we’re going back to some pre-SOA software engineering practices. I am pretty sure even Anne would actually with that (and toi be fair, she never said that).
2 Service-Oriented Architecture mobile edition // Jan 31, 2009 at 1:40 pm
[...] traded companies. The financial departments of publicly traded companies, that is. Gartner’s Nick Gall was kind enough to respond to my recent post with a clarification, in which he was quoted as [...]
3 all bits considered // Sep 1, 2009 at 6:37 pm
[...] Anne Thomas Manes article), dug up (for instance, the very same Joe McKendrick ) , and resurrected (You’ll all be doing SOA in 18 months ) Strangely enough, all these events appear to happen in no particular order… For me, SOA was [...]
Leave a Comment