It's been a busy year for IBM, from the acquisition of Cognos to the release of numerous products. From his seat in IBM's cockpit, Steve Mills has played an important role as his company developed and expanded on its Information On Demand strategy. IOD-a combination of data management, analytics, and industry-specific applications and services-is designed to help businesses more effectively leverage their data. With 2008 coming to a close, Mills, senior vice president of IBM software, discussed his views on IOD, cloud computing and SAAS (software as a service) with eWEEK Staff Writer Brian Prince.
People talk about IBM not being in the application business. Isn't that sort of a fine line with everything IBM does and the requirements of IOD?
The truth is it's always been a fine line. Our position has been one of building an ecosystem and not competing with our partners.
So you know Oracle has their broadly defined database and middleware product set, which obviously competes with us directly, and then Oracle has a whole set of application assets that are the various and sundry companies they have acquired, focused on CRM and ERP and industry-specialized application function. Those are areas where we're not competing, and the fact of the matter is we partner appropriately with Oracle, with SAP and with thousands of other application vendors.
That ability to broadly build out an ecosystem of companies that want to partner with IBM is facilitated by the fact that we've kept our focus around this set of middleware capabilities and not gone into the core application spaces that the overwhelming majority of application vendors cover today.
How does IBM leverage IOD internally? How does it manifest itself inside IBM?
Well, we do federate a lot of data inside of IBM. So we have multiple data sources, data stores of different kinds that keep data in them. We then access that data, merge it and analyze it.
A lot of the reports and analysis that exist within the company are multisourced in terms of the data stores from which the information is extracted. It's then brought together in various types of analysis, from simple spreadsheets right through more complex analytics. So we look very much like what we tell our customers to do, [which] is leave your data where it is, and then pull it together for whatever the purposes are that you may have from a use perspective.
Where do you typically find that enterprises are on this journey toward what IBM calls an Information Agenda, this information-based enterprise? How far along do you think the typical enterprise is in that journey?
I think that most of them are at the early stages, which is not to say that they haven't invested in integration technology and that they don't try to move data around, consolidate data today. They're all doing some of it.
For large businesses, you wouldn't say that any of them have yet to get started. They've gotten started. But they're very much in the early phases. Most of their work over the last decade has been more focused on just simply trying to interface the systems that are there [with] each other, rather than create a full integration architecture.
Now, I would say with the acceleration of [SOA] service-oriented architecture models of process integration that really began to build momentum in 2004, that post-2004 there's been a big increase in the interest in data integration as a companion to process integration. Customers that see the benefits of horizontal process integration creating integrated order-to-cash models straight through processing, as soon as they get into that on the business process side they immediately discover that they have to start thinking about common data architectures and data integration. And that's helped accelerate our whole information integration business over the last four years or so.
But they still have a long way to go. They'll tell you that they still have a lot of clutter, redundancy. They're still trying to create common data models for their businesses. So it's a long journey. My perspective is that the opportunity here, as we will experience it, is going to stretch out over the next couple of decades. There's that much work to be done to achieve a reasonable level of integrated data, consistent data, common data models, effective data mapping, data analysis. There's 20 years or so of work. I mean, there's at least 30 years of mess. You can't clean up 30 years in just a few.