Greg Todd is executive director of Accenture's information management services. SearchSOA.com's Rob Barry interviewed Greg about how MDM fits with SOA and BPM. You can download this and other interviews about SOA, BPM, and MDM as a podcast.
Rob Barry, SearchSOA.com: So first off I wanted to ask how MDM fits in with SOA and BPM?
Greg Todd, Accenture: I think what we've been finding recently is that BPM and SOA are tools that get enhanced when you use MDM as a centralized repository. So right now we are seeing more of a move of organizations centralizing master data and using SOA and BPM as the essential tools to integrate that master data to a number of disparate systems.
SearchSOA: And do you see a movement away from the more batch-oriented style of MDM?
Todd: I would say in some cases. I deal with a lot of large global Fortune 500 companies and in most cases there are still batch updates to the master data systems and there are SOA-based types of updates going on as well. So really it depends. In the large global companies you'll probably see more of a mix. But ultimately you'd love to get away from the batch loads, especially for global organizations.
SearchSOA: Now how about the frequency of updates, is that a tricky point in new MDM deployments?
Todd: It's interesting. It really is dependent on how the data is going to be leveraged during the day. So frequency could mean anything from let's update it once a week to once a day to once an hour to more of a near-real-time type of update. And it really is dependent on how critical that information is needed for exercising some sort of business activity. So, from a reporting perspective, most data is usually several hours if not days old. So therefore the master data is not that necessary for being up to date exactly at that same time. However there are new applications that are starting to come out where we're seeing the need for master data to be more up to speed with the analysis that's being done. This usually applies to more predictive analytics or some analytical type of capability, say, an optimization engine is running. SOA helps to enable the fluidity of master data in those types of analytical systems.
SearchSOA: How significant of an issue is speed in MDM?
Todd: When you talk about speed I think about how many records we're updating and how often we need to make those updates. I think that, given the technology and where it is today, speed is not really an issue. You can go as fast or as slow as needed. Again, most companies are just trying to sort out when they need to use that information and what types of critical business decisions are being made on that information. And that usually dictates how fast they need to have that type of information updated. So I think speed is not really an issue with the environments we're seeing for the large global companies. And I think when you talk about cloud computing and where the architectures are moving to in the future, it's going to be another opportunity to increase the speed and the frequency of updates to MDM tools.
SearchSOA: And of course the third piece when it comes to metrics is volume. Do companies sometimes try to pack too much volume through their MDM systems?
Todd: In most cases I would say yes. I think that the breadth of a record versus the counts for the number of records is a challenge in MDM. And what happens I think is that we try to put as many characteristics as we can for an individual record so that we have full understanding and full knowledge of everything that's going on with that record. So I would say yes, in some cases the breadth of the records are larger than necessary.
SearchSOA: Now how do you handle situations where, for example, different departments in a company might all want their own definition of a customer reflected in the master data sets?
Todd: This is a common occurrence. What we find is that there are individuals within organizations who would prefer to have ownership of their master data sets and we usually address this through a very rigorous master data governance and SOA governance program. What we do is set up, from a sustainability perspective, an organizational mindset around how data is going to be moved and what data is going to be moved. So all those comments before around frequency, speed and volume—we can add the word definition to that list. The enablement of an SOA or MDM-based governance group is usually what's called for in order to mediate the differences [of definitions] between departments.
SearchSOA: How about during mergers and acquisitions—how difficult can it be to integrate two completely different MDM systems?
Todd: We're seeing a lot of this, especially in the financial services industry now. There's a lot of M&A activity that's going on. And it's a challenge; it definitely is. I think technically the challenge is to again understand where all the repositories are of what key master data that organizations want to maintain. The combination of customer files and vendor files and things like that are a bit challenging because you have to match the records and do some sort of translation from one system to another. But on the flip side, although it's complex and challenging during an acquisition to do that type of work, that is the most opportune time for an organization to take a step back and really plan and do a good strategy around how they want to manage master data in the future. Sometimes an acquisition is a catalyst for organizations to build stronger master data management and SOA-based solutions.
SearchSOA: And what happens if they don't?
Todd: I think you'll find that organizations will fail in their need to report out accurately how well the acquisition is doing. So what will happen is, if the right customer files aren't updated in the appropriate ways and the information isn't available for the executives to review and drive the metrics around an acquisition, you'll start to see a little bit of frustration that you can't report on how the metrics around that acquisition.
SearchSOA: And finally Greg, can you offer any best practices for others who might just be starting to look at MDM in the context of BPM and SOA at their enterprises?
Todd: I think MDM is really a foundational capability for SOA and BPM. I think what we've seen is that over time organizations have started to mature their capabilities around building enterprise service buses and starting to integrate a lot of decentralized environments and starting to come toward a more centralized approach to how they manage their IT infrastructure. Being able to do that with their MDM is as critical today as ever given the large volumes of data that are available to anyone. Having that centralized architecture for MDM – having the ability to harmonize that batch to all the applications that need that master data using your SOA architecture – leads to more of a globalization of your master data management. So having it harmonized, synchronized, centralized and then globalized is the direction I think we're seeing most organizations heading.
This was first published in April 2010