Wednesday, April 23, 2008

Reading notes for "Universal Metadata Models" Part I

This is the second book by David Marco on metadata management in large enterprises. Michael Jennings is the second author of this book.



Part I, Presenting the Managed Metadata Environment

First, a managed metadata environment (MME) is far from being just a repository of metadata or a data warehouse. Building a data warehouse for metadata is much more difficult than a data warehouse for normal data because of the various data sources and the difficulty of maintaining the link among the data.

There are companies trying to do point-to-point integration with EAI tools or just using XML. What can be problematic with such approaches is that, when the company grows too large, the effort to maintaining the integration will very easily over the whole IT budget. So, it’s better to start thinking of building an MME as early as you can.

It’s quite inevitable that, when thinking of building MME from an architectural point of view, there are architecture elements that are much similar to what data warehouse architecture includes. Specifically, they are, sourcing layer, integration layer, repository, data management layer, data marts, and data delivery layer.

Similar to the development of a data warehouse, the governance and stewardship models of metadata are also important.


What can we achieve when we have a good MME? If we look at the SE-CMM for data warehousing (CMM is always something big enterprise is focused on and it indeed can help these enterprises to improve the business by reducing cost and promoting new business opportunities), most global 2000 companies are at levels 1 or 2. By having a good MME, it is possible to move up to level 3 and level 4 (it’s called “world class”). And of course, level 5 (continuously improving) has never been achieved by anyone yet.



No comments: