Related Links

Web Seminars

What trends are you seeing in organizational models around data integration?
For a long time we've seen organizations that are very fragmented, with different tools and different teams doing various kinds of integration in different ways. We're making the case and starting to see traction around a trend of rationalization of tools or standards. You can call it a competency center or whatever you want, but it's really a trend toward centralization of skills for managing and integrating data.

Yet there are all kinds of different integration scenarios across the organization with different requirements.
Yes, and that's the point. You don't need three ETL [extract, transform and load] tools, you need one. You don't need three ways to do data federation, you need one. You want to standardize the portfolio of tools and the process by which we design and develop, govern, monitor support and manage change. For example, there is a shared services team at Pfizer for BI and data integration that has taken out millions in cost through centralization of infrastructure and standardization on tools. They have a concept they call the playbook that's really a set of reference architectures. As new products come down the pike they map it into one of the plays which specifies what tools to use and what the architecture blueprint looks like. It lets project teams build out what they need rapidly and with greater quality over time. It's a classic example of the shared services idea.

Is there a risk that standardization will hold back those business units that had already optimized their work on a project basis?
A properly funded model should allow the business to act on projects fairly rapidly, but you raise a good point. If the funding is too centralized, you can end up with the old IT bottlenecks. You want something in the middle where the model is somewhat centrally funded, but then budget the business units leveraging those activities so they will pay for the work they need to get done.

Yet we've seen cases where the organization driving the standard is at the head of the line when it comes to benefits, and we've seen that other units will lose efficiencies.
I think in the case of data integration tools there can be win-win outcomes. Improving quality, reuse and better leveraging of investments can all be accomplished. But I agree that people are emotional and we all have baggage, we're married to the tools that we know and use. In that sense, people are going to be forced to give up what they know, which will be a painful thing. In our view, though, the benefits of standardization are self-evident.

Some Gartner surveys say that business users want to innovate on their own desktops and that they've lost faith in IT generally.
A flip side about IT's lack of willingness to let go of control has to be married back to the governance process. I fully support the idea of giving business users more power to integrate and organize data the way they want it. The thing is that they are pulling things off the Web and who knows where else, and we need to quality ensure that. I think a competency center or governance team is where the business and this problem can connect. But yes, most IT people still can't get comfortable with the idea of giving a business user empowerment to pull data from here and there and mash it up however they want. That's a tough thing for them to get over, but the fact of the matter is their users are already doing it, so they'd better get over it.

Jim Ericson is editorial director of  Information Management, a SourceMedia publication. You can reach him at Jim.Ericson@sourcemedia.com. Follow him on Twitter at @jimericson.