Thursday, January 03, 2008

MDM - Part 1 - An Introduction

In continuance with my recent post on "MDM War", I would like to take you all into this enchanting world of MDM with a brief introduction.



In my own words , MDM (Master Data Management) is the single place where any kind of reference data would be maintained globally for an organization. All transactions and business processes would lookup to the services of MDM for their operations. Some of the important tpes of master data , that an organization would be maintaining are



  • Product

  • Employee

  • Customer

  • Location

  • Supplier/Vendor

As you can see, these entities can standalone and are independent of the business processes, that an organization would participate in. MDM allows the companies to consolidate the master objects which might be residing in silos, harmonize, enrich, and federate one common view of the organization's data to the businesses seamlessly. MDM, as misunderstood , is not THE TOOL which will do this magic. It still relies on people and processes to solve the puzzle. It provides the framework to achieve it without much fuss.


MDM is one application for the organization and not one for each business unit, though some of the services might be business-unit wise. For example, HR department wouldn't be interested in the Product Data and Sales Department wouldn't be much keen in the Employee's salary.

The supply chain in the picture gives a better example of how the different businesses in an organization would like to view the master data (Click the picture for better clarity)



In a nutshell, the key capabilities of an MDM tool are



  1. Master Data Integration

  2. Master Data Consolidation

  3. Master Data Quality Validation

  4. Master Data Enrichment (optional)

  5. Work flow based Data Maintenance

  6. Master Data Publishing


Any tool, which doesn't provide these features fails to provided a complete MDM suite. And one important thing, MDM has nothing to do with data warehousing.

4 comments:

Anonymous said...

Senthil, when you talk about Master Data Quality Validation, what do you actually mean? Can you provide examples?

By the way, I am a product manager at Exeros- we do MDM validation between source and master data to make sure that data between un-retired source and master is consistent.

Thanks
Saravanan

Senthilkumar Bala said...

Yes Saravanan,

Master Data Quality Validation includes validating

1. Referential Integrity constraints
2. Handling duplicate records
3. Workflow mechanisms to publish invalid state of the records
4. Conformance to business rules

I dont quite understand when you say MDM validation between source and master data.

Anonymous said...

Senthil, I agree with all the vaildation you have mentioned. In addition to it, we need to do one more validation. i.e, to check if data in master is consistent with the data in source from which data was moved. Here's an example of what I mean...

If source A has address represented in multiple fields... Address 1, Address 2, City, Zip, Country, but master data may just have one field called address. So the relationship is
masterdata.address = sourceA.address1 + sourceA.address2 + sourceA.city + sourceA.zip

when data is moved from sourceA to master, we need to make sure that the actual results are the expected results which is one important piece of master data validation.

In this example, after a record called

Senthil, first street, bangalore, india is moved to master, we should validate and make sure that the data actually moved properly and the resulting master data values match the source.

By the way, your MDM war article is interesting, I am looking forward for your next article on it.

Senthilkumar Bala said...

Saravanan,

I read about Exeros MDM Module. Its very interesting. Is it possible for me to contact you on the tool ?

regards,
Senthil