I've been meaning to catch up with Oracle to get an update on its AIA because the concept was upgraded to version 2.5 at Oracle OpenWorld in October. But the task kept falling down my to-do list. None of you are interested in enterprise application integration (EAI) anymore, I figured.
But a November post here on IT Business Edge by Mike Vizard (see Composite Applications to Define Enterprise Software) got this blind guy to walk around to the front of the elephant and take another feel. Mike looked at Oracle AIA and I believe what Mike is calling the post-enterprise-application era is what I call business process management (BPM). As in "<strong>BPM is the next big thing in ERP</strong>?"
BPM is all about tying business process sets together efficiently and effectively, and most business process sets are currently automated by disparate packaged and inhouse-developed software. You can tie them together in two ways:
Clearly, years ago you in IT decided you would take the second option and that led to EAI. But EAI is cumbersome to use, highly point-to-point in orientation, and expensive to implement and reimplement when business processes change, as they always do (mostly because of the first two characteristics in this list). BPM takes EAI up a level of abstraction, hopefully reducing costs because it puts more of the implementation and reimplementation work on the users of the software and less on IT to handle change. I am not going to go down the BPEL/workflow/system-to-system rathole here. I don't pretend that the ideal BPM package exists yet. But it's all BPM to me. And should be to you.
Which is why AIA is so interesting. Oracle is the perfect test bed to see if the ideal BPM package can be built. No one has more disparate packaged software in one company than Oracle with its dozens of acquisitions over the last 10 years. But it leads to the recurring question: Do I want to get my EAI/BPM from the same vendor from which I get my enterprise applications? Or is it better to look at a standalone BPM supplier (there are only a few left) to avoid vendor lock in? It's the old stack guys vs. pureplay discussion except looking from the top of the stack instead of from the middle or the bottom.
(As an aside, I disagree with Mike Vizard's concern that enterprise application modules or components or process sets are becoming commoditized and therefore not adding any competitive value to your enterprise. That's what you want because it holds down costs. What's the competitive value in how you process the payroll on Thursdays? But I can save that disagreement for another blog post.)