Can the Web teach business IT-even enterprise IT-about integration?
Traditionally, the logic has been that Web technologies and approaches are great for the Web-but not for internal, distributed systems.
But Dr. Jim Webber, global head of architecture for ThoughtWorks, seems dedicated to proving otherwise.
Last year, I shared with you a presentation Webber made at Qcon about the Web as middleware and why you don't need a "bloated" ESB. This year, Webber returned to Qcon, this time with a new speaking partner - Ian Robinson, a ThoughtWorks principal consultant who specializes in the design and delivery of service-oriented and distributed systems.
The two are writing a book on Web-friendly enterprise integration. During this year's Qcon, Webber and Robinson sat down to a Q&A session about using Web approaches for enterprise integration, which InfoQ posted as a video-with, thank goodness, a transcript - last Friday.
Webber, as always, is entertaining, with his quick British wit. Robinson brings his own humor, plus a knack for explaining really complicated topics. However, I'd be derelict of duty if I didn't tell you the discussion leans heavily toward the technical side of the equation, despite Webber and Robinson's best efforts to keep it understandable and simple.
Much of the interview focuses on REST, although Webber and Robinson stressed they're not from the "Church of Rest" and did look at how other Web approaches could be used to integrate enterprise systems.
What's interesting to me is that I'm 95 percent sure Webber and Robinson aren't talking simple integration-by portal, say, or API. They have backgrounds in distributed systems integration, and Webber specifically talks about replacing enterprise middleware with what he calls "webby" approaches:
"Indeed, my current clients, who are very interested in massively high performance systems, would have potentially gone with traditional enterprise middleware if we haven't done some empirical experimentation and found that a simpler webby approach was actually just as well suited for their needs. Having done those empirical data points it sort of emboldens you a little - you kind of figured out that the web stuff works quite well once in one scenario, and then again in another scenario, and it embolds you to default to that when you are talking about distributed systems integration."
If you're interested in the technical discussions, great-enjoy the whole interview. If you'd like to cut to the chase about what REST means for business, when to use REST and why you should avoid REST "platforms" and "boxed REST," then skip ahead to questions seven, nine and sixteen. You'll need to count the arrows in the transcript, but when you click on the question, the video will jump to the relevant part. Or you can just read the transcript, which is faster, although less entertaining than viewing the video.
Since we're on the topic of what the Web can teach enterprises about integration, you might also want to check out "Running your SOA like a Web startup," an excellent post by ZDNet blogger and IT consultant Dion Hinchcliffe. He looks at open APIs, which he also calls "public-facing Web services," and suggests organizations apply the success lessons of open APIs to SOA.
It's a complex discussion, which I won't attempt to summarize here, but this blurb contains the core theme:
"It's very telling that Twitter still gets 10 times more usage through its API from partners than through its Web interface and it not unusual. It's also very uncommon for a new Web product to launch without an API; services are the route to success in today's Web where if you have anything of valuable, you need to make sure it has leverage. SOA does not have the same business urgency and lacks critical focus in this regard in most organizations. In short, traditional enterprise SOA has a lot to learn from the open API world."
And, if you ask Webber and Robinson, traditional enterprises also have a lot to learn from the WWW.
It's not a new discussion-Anne Thomas Manes made similar points in 2007 and you may remember the WOA discussion-but it does seem to be a topic analysts and experts are taking more seriously. Hopefully, we'll hear more about it in in the near future. I would expect the release of Webber and Robinson's book to generate more discussion.