In the latest issue of Information Management dropping on desks next week, you'll see an article on the merging of business intelligence and business analytics, and an argument that calls for an integrated approach. It asks traditional vendors to embrace advanced predictive, data mining, text and other advanced capabilities in their business intelligence platforms.

Vendor economics will dictate whether this becomes reality, and you'll have to read the article yourself before deciding what you think. I'm not here to judge, but the author's message already phoned home this week in news of a little feuding between SAP and Oracle, after the latter released Oracle financial analytics for SAP applications.

In other words, Oracle says, let us help you do a better job understanding the info in your SAP applications.

SAP, as you'd expect, replied to say its financial analytics and analytic appliances are just fine, thank you, and already more popular than their competitors' in customer's eyes.

What I was left thinking was, if there are marketable arguments on either side, is there value in standardizing on analytics from any vendor, especially with so many use cases out there? Why not standalone or analytic service providers? What's so hard about applying analytics where you need them -- operationally in alerts and dashboards, visually in depictions, in business rules or in enterprise reporting -- and why would an ERP vendor think it was a good idea to jump into that market?    

This is a longstanding discussion in our community that usually draws on the fact that BI and analytics are dealt with separately, or suggests that analytics are the "new BI" or whether embedded analytics or a standalone analytics layer belongs in our technology stack.

The upcoming article in our magazine takes a position but I am not sure any of the constructs above are entirely true or false. I'd expect your point of view more likely depends on what you are trying to do and the interests you are defending.

I was talking with Boris Evelson at Forrester about this the other day, and he reminded me that on one hand, merely embedding analytics in siloed apps sounds a lot like the software excesses of the 1990s and beyond. In a real enterprise BI architecture, the challenge is much more in the basic components of data quality, ETL and/or MDM that lay a foundation for a centralized analytic approach that might follow, he says.

That makes perfect sense to me, and while we've created plenty of "real" enterprise architectures to attack the heavy lifting of ad hoc BI, we also know how these have mostly performed (not very well) against demands for reasonable timelines and dynamics that change with every new business event or leadership regime.

The Oracle/SAP spat really boils down to a technical question of whether Oracle or SAP is better at reading SAP metadata, Evelson figures. "SAP and Oracle and IBM and Microsoft are all open architectures. It's not whether Oracle can embed in SAP apps, the question is whether Oracle understands SAP metadata as well as SAP does, that's what you'd want to ask."

That's an elongated technical argument you would want to ask if it's your job to know such things. But that doesn't address all the unsorted managerial EDW challenges, and makes it no surprise that operational analytics, in-memory analytics and maybe even mobile analytics look so desirable right now for a quick means to a limited but important goal. Analysis, human, automated or both, is in great demand constantly at every useful level of complexity.

Even the "idiot lights" in cars that monitor voltage or oil pressure are more likely to be computationally driven today. I'm frankly threatened by the thought of what an iPad is going to offer to guide the course of my day sometime in the future.

At a richer level, light but powerful analytic tools like Tableau or Qlikview seem more about visualization than BI reporting, but I don't want to get in trouble with the semantics of what they are, another reason not to nail down architecture too firmly. Such tools are uncomfortable for traditionalists, Boris and I agreed, because in-memory analytics can undermine the control mentality of a single version of truth and carefully managed infrastructure. (From a user view, who wouldn't want the BI equivalent of free, front of the line access to all the rides at Disneyland?)

Back at enterprise scale, the ERP solutions also have to contend with big power tools of a SAS, the proven legacy of a Microstrategy and another multitude of smaller specialists who are not wisely underestimated. Because of their sophistication and complexity, all these products fall more comfortably into the control culture of enterprise architecture. I doubt many folks are convinced yet that it will be sorted out with big BI vendors leading the way (though they will be buying smaller maturing competitors). 

So the stack marches on with analytics of every flavor poking their way in from simple business rules to portfolio risk hedging. The simple ones, like analytic idiot lights in cars, we can carry with us. That great big Lexus diagnostic computer is probably better left in the dealership where it can serve a lot of cars with more complicated problems.

Well, at least until they dumb it down for casual users and iTunes comes up with an app to access that front end too.