Columns

Data warehousing industry weaves a meta data standard

• By David Marco
• May 1, 2001

Companies have realized the key to their prosperity is how well they gather, retain and disseminate knowledge. And many of these same companies understand that meta data is the key to these processes. For meta data to be valuable to a corporation, it needs to be extracted from various software tools and integrated into a meta data repository. Rarely do companies purchase the majority of their software tools from the same vendor. This best-of-breed approach allows companies to select the tools that best fit their needs. On the other hand, it creates non-integrated products and makes the task of meta data integration difficult, to say the least. This issue of meta data integration is one of the chief mitigating factors that have prevented most organizations from achieving successful data warehouse, e-business, Customer Relationship Management (CRM) and Enterprise Resource Planning (ERP) implementations.

This column focuses on the Object Management Group (OMG) meta model standard Common Warehouse Metamodel (CWM), the impact this standard will have on the industry and its promise to aid in this task of meta data integration.

Companies that have built a meta data repository will typically extract technical meta data from many of their existing software tools. These commonly include front-end tools (such as Business Objects and Cognos), data modeling tools (such as Erwin and Silverrun), databases (such as SQL Server and DB2), and extraction, transformation and load tools (such as Informatica and Ascential Datastage), just to name a few. In an ideal world, there would be tools available that integrate all of a company's sources of meta data into one integrated and architected repository; however, that utopia just doesn't exist. Today's meta data integration architecture involves a series of proprietary vendor products attached to one another with chewing gum, string and paper clips.

Now before you give up hope, the OMG CWM is a standard that offers the promise of improving these meta data integration processes. But what is a meta model? It is a fancy phrase for a physical data model that stores meta data. The CWM has initially focused on the data warehousing arena and is broadly supported by the vast majority of data warehouse vendors, meaning that they have integrated CWM into their tools' meta model or they are looking to provide an interface that will transfer their meta data into CWM. This capability will allow data warehousing products from different vendors to share technical meta data. The CWM specification can be downloaded from www.omg.org.

For many years all of us in the meta data arena have desired a global meta model standard. A year ago we had two competing standards, CWM and the Open Information Model (OIM), which was being moved forward by the Meta Data Coalition (MDC). Unfortunately for the industry, two standards were one too many. On September 25, 2000 the MDC merged with the OMG with the goal of consolidating the separate initiatives into one meta data standard under which all vendors can unify.

At long last, we have a meta data standard in the industry: CWM. Software vendors need to get to work integrating it into their tools, because it is what their customers want. Now the question is: Once vendors standardize on this model, will they be able to share 100% of their meta data with other tools? In theory, yes. But I believe that 60% to 70% is much more realistic. Both software vendors and system integrators should utilize this standard.

Most corporations will not purchase a tool to build their meta data repository, but will choose a custom implementation. If vendors integrate their tools on CWM, corporations should not wait for this to happen. A meta data repository can provide your company a strategic advantage in the marketplace today. So don't wait for the vendors. Move forward and you'll get fantastic results.