Content Copyright © 2005 Bloor. All Rights Reserved.
The IBM Information Management group has been taken over by Ambuj Goyal following the retirement (actually she’s not gone yet but will be doing so shortly) of Janet Perna. As a result, he and his team have been busy briefing analysts as to the company’s plans, particularly at a recent event in Boston.
Given the breadth of the Information Management group it will come as no surprise that there was a lot of stuff in which I was interested, most of which I do not have the space to consider here. However, there are a number of things (actually, one thing, since they are all related) that came out of the meeting that are worth discussing.
The overall strategy that IBM was expounding was what the company calls Information as a Service which is, in effect, how the Information Group supports the whole IBM Information on Demand strategy.
In what way should we think of information as a service? Perhaps the classic example is a call centre application which might want to be fed, for example, with live data extracted using WebSphere Information Integrator. However, IBM expects to see the emergence of new applications in a much broader category than this. First, it sees a growing requirement for applications that combine data and content, which is why DB2 is being extended to include native storage for XML and, secondly, it believes that master data management has the potential to change the face of development.
This is worth exploring further. IBM’s vision is that master data management (MDM) is owned by the enterprise rather than lines of business and that the master data management system is referenced by other applications as and when required. In other words, data is provided to front-end applications from the MDM system as a service. To put this another way, instead of a three tier architecture in which you have the presentation layer, the business logic layer and the data layer, you now have a four tier architecture in which MDM sits between the business logic and data layers.
In effect, this means that all the definition of data is done within the MDM systems, which means that you don’t define it in your applications—leaving applications just to worry about business logic. This has a number of implications in terms of what application/integration development environments will need to provide (both less and more) but that is a discussion for another day. What it certainly does mean is that MDM will be absolutely fundamental to enterprise software.
It also raises another interesting question: if MDM is implemented on top of DB2, say, then how much of what is in the MDM can be automated as a part of the database layer without requiring specialist MDM capability? This is a question I asked IBM but it was unable to answer. Dan Druker, who is director of enterprise master data solutions within the Information Group, conceded that this was a possibility for the future but that they hadn’t got that far yet. That’s a fair answer: MDM is still in its infancy and automating some of its functionality into the database is a long way off.
I think that the concept of information as a service, especially when linked to MDM, makes a lot of sense. I think it will take a while but it must make more sense to consolidate on a single definition of master data than the current situation in which multiple versions of the truth proliferate and then we have to go through all the hassle of integrating and cleansing them. This also suggests a split in the MDM market: products that are designed to get over the problems that currently exist and products that are designed to prevent those problems occurring in the first place. The latter may be a longer term goal but it is one that clearly makes sense.