I'm intrigued by the OMG's Common Warehouse Metamodel. In what areas is this model lacking? What are its main pitfalls and downsides?


Mike Jennings' Answer: The Common Warehouse Metamodel (CWM) is an open standard for the interchange of data warehouse-specific meta data focused on being vendor neutral. Meta data interoperability, across applications, is accomplished through the use of a shared meta model. The Object Management Group (OMG), www.omg.org, a consortium of more than 800 member companies, governs CWM. OMG is known for the Common Object Request Broker Architecture (CORBA) standard. The current version of the CWM specification was released in June 2000.

CWM borrows some of its designs from the Open Information Model (OIM), which was from the Meta Data Coalition (MDC), originally was designed by Microsoft and Texas Instruments in 1996. In 2001, the MDC OIM was officially combined their individual efforts into the OMG. A key point about CWM is that it was meant to be specific to the data warehouse and business intelligence meta data focus area, versus OIM, which also covered business rules and business processes.

Integrating meta data provided by different software products (e.g., ETL, OLAP, data profiling, data quality, database, etc.) is complex, since each tool has its own proprietary design technical and business meta data storage. The CWM is a model for data warehouse and business intelligence meta data. It is intended to be used by vendors of data warehouse and business intelligence software so that the products can interoperate with each other.

Most data warehousing meta data integration efforts require a significant amount of time and effort to implement. CWM acceptance as a meta data standard in the various DW products hopes to reduce this high-level cost and to increase the effectiveness meta data management. As CWM gains wider adoption in the industry, the data warehousing industry as a whole should benefit from the use of the standard and improve implementation projects.

CWM only covers data warehouse and business intelligence meta data, but it has other limitations. CWM is not a repository standard; it only specifies the shared models along with some supporting underlying technologies. It does not specify any repository functions that would be needed to support the operational needs of an enterprise meta data repository project.

A future direction for CWM is to utilize Web Services to support distributed meta data architectures. Another opportunity is defining the levels of interchange or granularity between the various data warehouse products to optimize interoperability through using the standard.

Evan Levy's Answer: The key to success in delivering an enterprise data warehouse to business users is ensuring business action is possible from the data and analytical functions delivered by the warehouse. Bearing that in mind, there are a few functions that I believe are important to your EDW center.

The business requirements should expand beyond the traditional function definition activities typically performed by IT development teams and include data requirements and functional requirements in addition to identifying business requirements. Business requirements should include the business questions, the business actions, and the business situations and scenarios (this ensures that timeliness, delivery methods and other details are identified). Data requirements are the first step in supporting data standards (that you identified as a center responsibility above). It should include the identification end-user business names, definitions and domain details (from a business-user perspective). The functional requirements will identify the details that the development staff requires for implementation (data delivery, presentation, availability, etc.)

The statistical reporting area should focus on end-user data delivery. While you've identified statistical analysis, don't disregard the value of delivering standard reports and simplified ad hoc analysis. The majority of business users typically want standardized reports to allow them to focus on their jobs (rather than learning technology and data navigation). You might want to consider providing reporting and end-user development support initially to ensure that the EDW data is making its way to end users. Supporting business actions is crucial - not every business action will require advanced or complex analytics. More often, a simple canned report is sufficient to support an end user.

Your data standards area should include data administration, data modeling and meta data management. The data administration function is very important; it should include developing and documenting the data content asset (from terminology to definition to domain representation). Data administration should also focus on identifying data-level business sponsors as well as source system data stewards. These folks will be important in implementing data management functions (data quality assessment, data correction and subject/content expansion). Meta data management and data modeling will work together to ensure that the data administration content is developed and packaged for end user and application developer usage. These two roles aren't complex - it's just important that they are established on day one.

The EDW management function should focus on ensuring the EDW project is maintaining business value and focus. Many organizations decide to have project management and administration reporting to this area. We believe it's important that the EDW management area also focus on end-user expectation management - this includes tracking budget and business benefits, ROI calculations, tracking new business initiatives and changes, and participating in the business requirements process. EDW management is different from traditional IT program management because the business value metrics are different. Operational systems are managed for processing efficiencies; EDW systems are focused on enabling business decisions and the business value associated with those actions.

Register or login for access to this item and much more

All Information Management content is archived after seven days.

Community members receive:
  • All recent and archived articles
  • Conference offers and updates
  • A full menu of enewsletter options
  • Web seminars, white papers, ebooks

Don't have an account? Register for Free Unlimited Access