MDM - a strategy for IBM
'You're on the right track, but...'
Comment This is the second article of two stemming from IBM's recent European MDM (master data management) conference. In the first article I discussed how I disagreed (at least in part) with the view of MDM that was put across at that conference. This article concentrates specifically on what IBM is doing in this space.
In Harriet Fryman's recently published MDM Report, IBM did not score highly. However, my inclination is to think it will score much better in the future, because the company's strategy appears to be the right one, at least in so far as it intends to move away from a siloed approach with separate emphases on products, customers, suppliers and whatever, towards a more broadly-based platform that supports non-domain specific master data management.
IBM will score highly because it is likely to be the first company to introduce this sort of MDM platform. I understand that a number of the specialist vendors are trying to extend their existing products but are having some difficulty in achieving this—people data and "thing" data are significantly different (IBM's view is that location data is also different, though I am not convinced on this count: location is, after all, a thing with a different set of attributes).
Of the other major vendors, Oracle has issues to resolve with respect to Oracle versus Siebel, before it can start to put together a platform and SAP appears to be behind the curve. Other major players that may enter the market like Informatica, Business Objects, Microsoft and Sybase, have even further to go.
One other thing I think IBM should do is to stop focusing on its products purely as hubs. Yes, some companies want hubs. And, yes, there are lots of bucks in hubs. But there are also lots of companies that would be better suited by a registry or repository-based approach, possibly with the intention of migrating to a hub later. You can, in fact, implement IBM's products in this manner but it is hardly mentioned in corporate marketing and nor is there any detail about how you would move from the federated model of a registry/repository to the persistent model of a hub (the services interfaces would be the same).
Another thing that IBM is well suited for is to create synergies between MDM and CMDB (configuration management databases) about which I wrote here recently. In fact, I asked a number of IBM personnel at the conference about this and the responses, when I found people who understood what I was talking about, were illuminating.
The first thing I heard was that a number of customers have apparently asked the same question and the second thing was that, yes, they have been talking to the Tivoli guys. Of course, talk is cheap and they couldn't tell me anything about concrete plans but at least that is hopeful. Moreover, it would give IBM an edge over its competition in either the ITIL or MDM spaces because of the ability to redeploy the same technology for multiple purposes, which could well make investment decisions easier for users.
This raises perhaps the most interesting aspect of the conference: where do you sell MDM? MDM is essentially an IT function and that tends to be difficult to sell, which is why reuse as a CMDB would be useful and also why an iterative registry-repository-hub approach could be beneficial. To sell MDM to the business you need some sort of business case. One such might be data governance and IBM made a substantial play about MDM supporting data governance at the conference, including working with IBM Global Services to provide a framework and best practices for setting up data governance procedures and policies.
The other possibility is specific applications such as "new product introduction" or "enterprise customer search", which address specific business pain points. However, the problem with this application-based approach is that not all users will be able to think for themselves of potential applications, so it really needs IBM to suggest them.
Actually, this is more broadly true of IBM. It is also true of the XML/relational applications that IBM hopes users will build using the latest release of DB2, and it is true of Entity Analytic Solutions (where might its capabilities best be embedded?) and so on.
MDM is not a technology looking for a solution - it has clear benefits - but quantifying those benefits and putting them in business terms is not so easy. My message to IBM from this conference is that it is on the right track but that it needs to clarify both its technological strategy and its business value.
Copyright © 2006, IT-Analysis.com
Sponsored: DevOps and continuous delivery