Integration at the Glass and the 80/20 Point
Is "Integration at the Glass" such a bad thing or does it satisfy Tim Bray's 80/20 point?
"Integration at the glass" is a term associated with Portal development, where developers or users can quickly link portlets together creating composite applications. The contention that such a technique seems to fly against traditional development wisdom. For instance, Glenn Kelman, vice president of marketing and product development at Plumtree thinks the "integration at the glass" catch phrase is "somewhat derogatory" implying a more superficial kind of integration [source]. However, is it truly superficial or is the way people doing integration, that is either via a shared database or via messaging middleware, simply wrong?
One of, if not the biggest problem in most IT settings is the problem of integrating disparate applications. The traditional approaches have been either to embark into a massive campaign to refactor existing application to use a common data model and data source. Failing to ever achieve agreement on the former, the alternative is to employ message oriented middleware. That is, if we can't have a common database, we can at least agree on using a common communication bus. But, I'm wondering, is "integration at the glass" a viable and sound alternative to these 2 traditional forms of integration? Because if it is, then it would definitely shake up a lot of grand plans in a lot of corporations.
So, let's go about the reasoning as to why it may possibly be the correct approach. The argument has its underpinnings in Tim Bray's 80/20 point, which is shortly described as:
Technologies that hit the 80/20 point win, and those that don’t lose, and that’s all there is to it; it’s hard to think of significant positive or negative failures.
The fundamental difficulty of integration lies in Conway's Law and our human inability to achieve a common viewpoint. "Integration at the glass" appears to be a late-binding approach to integration, that is integration occurs at the last possible moment. That is when all the data has been transported to the same place and presented to the user. At the point in time when the user has discovered what he needs most to achieve a solution, something that reminds us of YAGNI. That is integration happens only when a end user needs it, not months or years before.
So if "integration at the glass" does 80% of the integration with 20% of the effort, does it not clearly satisfy the 80/20 point and therefore is bound for technology success? So maybe its time to disregard common wisdom, afterall all other 80/20 point technologies also been previous denounced as being mere "toys".
Last modified 2004-03-16 04:51 AM