01 Jun 2010
MDM 101
In the information management space, few technologies are growing as fast as Master Data Management. While IBM’s acquisition of Initiate Systems has shed additional light on MDM, it’s hardly ubiquitous. There are plenty of people that either:
- Don’t understand it
- Don’t understand the need for it
- Don’t understand its limitations
In this post, I’d like to illustrate why MDM matters, a few of its limitations, and why I believe that it’s here to stay.
First, a definition from Wikipedia is in order:
In computing, master data management (MDM) comprises a set of processes and tools that consistently defines and manages the non-transactional data entities of an organization (also called reference data). MDM has the objective of providing processes for collecting, aggregating, matching, consolidating, quality-assuring, persisting, and distributing such data throughout an organization to ensure consistency and control in the ongoing maintenance and application use of this information.
Many large organizations have long been hamstrung by redundant or mismatched data sitting in silos, making analyses of customers and employees much more difficult than necessary. Simple questions such as, “How many customers bought from us last quarter?” can easily become long, complicated exercises in data extraction, manipulation, and analysis.
A few more thoughts about MDM:
- Because it’s so new, relatively few organizations have purchased formal MDM applications.
- Many organizations have been using de facto or homegrown MDM solutions for years to combat these very problems. Translation or “XLAT” tables, ETL scripts, and other creative “back end” solutions come to mind.
Limitations of MDM
At its highest level, MDM addresses a fundamental challenge confronting many businesses: data in multiple systems often becomes splintered or redundant. While not comprehensive, a brief overview of MDM’s limitations is in order.
First, contrary to what many believe, the purchase and implementation of an MDM tool is not a silver bullet. Overly complicated architectures, cultures that ignore data quality, and poor corporate governance don’t magically disappear because of the introduction of an MDM tool.
Second, there no such a thing as a “golden record”, as MDM expert and author David Loshin has pointed out.
Finally, while this may be an overly simplistic view, MDM may be a bit of a “Band-Aid.” In other words, organizations may benefit more in the long term from the following:
- Utilizing simpler architecture.
- Moving towards a model of data governance as described in Tony Fisher’s excellent book The Data Asset.
- Conducing routine data quality and data profiling exercises.
- Retiring legacy systems.
Realistically speaking, however, many large organizations have for years built a hodgepodge of standalone systems designed to accomplish one objective or serve the needs of one department—not the entire organization. These aren’t going to be instantly dismantled, especially in a down economy. Even organizations with enterprise-wide apps and systems often have exceptions.
The Future of MDM
For a variety of reasons, I expect us to hear more about MDM in the coming months and years. I believe that it’s firmly moving to the right on the technology adoption lifecycle.
Consider two types of organizations:
- Type A – Those that only use products and technologies provided by their approved vendors.
- Type B – Those that use whatever products and technologies make sense.
For more conservative organizations (Type A), Initiate’s MDM offering now has IBM’s imprimatur. In a way, MDM is now more palatable to them. For more cash-strapped or more adventurous organizations (Type B), there are alternatives to dealing with large software vendors.
Most recently, in late January of 2010, Talend announced the release of an open source MDM offering. Then, in April, the company received an additional $8M USD in venture capital funding.
Outcomes from these moves are likely to include:
- Users are more likely to dabble in the basics of MDM by playing with this solution
- Some subset of those users will engage with Talend proper, and license the legitimate application (and this would compare/compete with other MDM offerings)
- Some third party vendors or startups will probably build on top of current OS MDM apps to create more feature-rich MDM solutions. Forks can be huge. Consider that WordPress started as a fork back in 2003.
Conclusion
For a technology to gain acceptance, it has to be both useful and affordable. In its brief history, the question with MDM has never been about the former, especially in large organizations. OS alternatives have certainly addressed the latter, although open source and “free” should never be confused. The future of MDM is bright.
What do you think?


June 1st, 2010 at 10:26 am
Phil, excellent article on MDM 101.
MDM several years ago was a buzz, and like you suggest, it is now becoming a more mainstream strategy where it is now accepted as a reasonable initiative like Data Warehousing or ERPs.
Vendors do a great job sharing what is important, and what makes them stand out but this can clearly cloud the real reason behind an MDM project. First off, define what is master data to the business. Nothing like having a round peg being pounded into a square hole. Next the business has to have a value statement to proceed. That includes internal selling, between all levels of involvement. Executives, Business, and Technical. Lastly, MDM has to succeed, quickly and often for it to be a worth while story to continue.
Thanks for sharing Phil. I also look forward to the months and years to come in this space.
June 1st, 2010 at 12:53 pm
Phil,
Thanks for the MDM primer. I have been using MDM solutions since 2002 when my former firm purchased what was called the Razza MDM server (which was purchased by Hyperion Corporation and now part of the Oracle Corporation BI solutions. Many firms understand the value add of having an MDM solution and this is why Informatica Corporation purchased Siperian Corporation MDM solution, Microsoft Corporation acquired Strature Corporation several years ago and now offers what is called Microsoft Master Data Services. I have had the opportunity to work with the Golden Sources MDM solution that provides reference data for the Financial Services.
June 1st, 2010 at 5:17 pm
[...] MDM 101 « Previous Entry Book Review: Methland by Nick RedingNext Entry » 90s Sites and Stickiness [...]
June 3rd, 2010 at 8:51 am
There is a difference between “Reference data” and “Master Data”
Master data is the unique identification of key business entities, particularly where those are shared between business units, businesses, etc.
Reference data is the classification of business data, eg types, roles. Reference data has many names, for example controlled value lists, code lists.
The Wikipedia definition is inadequate in this regard.
Master data management then, at its core, is about ensuring the same business entity can be located by a unique master identifier (or identifiers) across multiple data sets, and ensuring that to the best extent possible a business entity can be matched with its unique master identifier.
June 3rd, 2010 at 9:33 pm
Thanks for the comments, all. I look forward to discussing issues such as these on this site. Stay tuned.
June 5th, 2010 at 9:13 am
Hey Phil
Great article, but I have a nit Re: RefMasterData.
Like many other distinctions, the one that separates master data and reference data is artificial. Your definition of the difference was right on the mark, but there is no compelling reason in my world to split them up. A master data entity needs to identified, classified and associated (as necessary) to other equivalent master data records. ‘Cisco’ can be ‘Cisco Systems’ in the same MDM with no problem.
On the other hand, a reference entity needs to be identified, classified and associated (as necessary) to other equivalent reference data records: ‘customer’ can be ‘client’ in an RDM with no problem.
In other words, they are more alike than different – make them part of the same application…you can thank me later. :~)
John O’
June 8th, 2010 at 3:53 am
The most difficult thing in achieving MDM, aside from getting the controlling parties to play nice is in restructuring business, so for example a “CUSTOMER” is only created in one place, then migrated to the various other databases/applications. Getting to a single copy and/or pace to hold it would be nirvana. Getting a large Corporation to agree on a single set of Reference Data or coming up with a Translation layer to relate the different sources is trivial by comparison. Managing the processes going forward with a Corporation that is constantly in acquisition mode is an even greater challenge.
June 8th, 2010 at 6:55 am
Terry – I have a question…instead of restructuring business, and since the same “CUSTOMER” could theoretically show up in differently structured “SUPPLIER”, “CLIENT”, “PARTNER” and “LITIGANT” database(s) would it not be simpler to create a single organization and reference the multiple occurrences?
June 8th, 2010 at 7:21 pm
Terry-
Great comment. About a year ago, I dismissed the need for MDM precisely because of this:
Getting to a single copy and/or pace to hold it would be nirvana. Getting a large Corporation to agree on a single set of Reference Data or coming up with a Translation layer to relate the different sources is trivial by comparison. Managing the processes going forward with a Corporation that is constantly in acquisition mode is an even greater challenge.
I realize now that larger organizations (and even some smaller ones with which I have worked) will never get to this state of nirvana. For that reason, I’m much more bullish on MDM.
June 10th, 2010 at 12:05 am
MDM Framework demo available
http://www.wetterlands.com/MDMF%20Demo/butterfly%20portal.html
This framework pulls togeather all of the elements required for the MDM space.
June 14th, 2010 at 2:48 pm
The better the data, the better the decisions (hopefully). And I agree with your assessment of IBM’s purchase of Initiate Systems’ master data management platform… good move on their part to bring data management services of this type to the mainstream.
June 25th, 2010 at 5:37 am
Do you have(or anyone in your company have) any kind of tool comparisons document between ORACLE, IBM and Siperian MDM suite
Any help will be greatly appreciated.
Regards
Pranab