A special thanks to Mike Needham for his invaluable contribution to this month's column.

Today with so much emphasis being placed on empowering the business users to do more with less information technology (IT) help and enabling the IT organization to produce better results faster, it is no surprise that meta data has become one of the hottest issues of the day. In order to leverage all of the meta data from the disparate sources that exist in your organization, you need a way to gather it into a central location (repository). This is where meta data integration tools come into the picture. Utilizing these tools, you can integrate your enterprise meta data into the repository and make it available to all the different user groups in your company. But, how do you determine which tool is right for your organization? There is a wide variety of tools on the market that claim to solve all of your meta data problems; but, as we all know, the vendor hype may not be a reality. This column will help you to objectively evaluate meta data tools.

Meta Data Sources

Within any organization there are going to be different sources of meta data. Meta data integration tools can integrate meta data from three broad source categories, each of which requires varying levels of integration complexity. It is important to identify all sources of meta data that you need to integrate into the repository. Figure 1 lists various meta data integration tool vendors.

Vendor Meta Data Integration Tools
Computer Associates PLATINUM Repositiory Open Enterprise Edition (UNIX)
PLATINUM Repository MVS
Informix Business Solutions Meta Stage
Unisys Universal Repository
Allen Systems Group ASG Rochade
Oracle Oracle Repository

Figure 1: Meta Data Integration Tool Vendors

The three classifications of meta data sources are certified, generic and non-supported sources.1 A certified source is a source that the meta data integration tool can directly read, properly interpret and load into the correct attributes of the meta model without changing the model. An example of a certified source is a CASE (computer-aided software engineering) tool such as ERwin. Most meta data integration tools are certified for several vendor tools. A generic source is a source in a common format (i.e., tab delimited, space delimited or comma delimited) that the tool can read, cannot interpret and may require a meta model change. Most tools support one or more generic meta data sources. An example of a generic meta data source is data that is stored in databases and spreadsheets. The last source of meta data is the non-supported source which is neither certified nor generic and may require extensive analysis to produce an in-house solution to access. The non- supported meta data source cannot be read or interpreted by the meta data integration tool.

Once you have classified the meta data sources available, you will be able to quickly determine the complexity of your project and if the tools you're evaluating support all of your sources. Figure 2 lists some of the more common meta data sources and how they are typically classified. (Your specific implementation may differ.)

Meta Data Sources Meta Data Description Type Model Extension
CASE Tool Physical and logical models, domain values, technical entity definitions and technical attribute definitions Certified No
Extraction/Transformation Tool Technical transformation rules Certified No
Custom Data Dictionary Business attribute and entity definitions Non-Supported No
Microsoft Excel Data steward's list Generic Yes
Reporting Tool Access patterns and frequency of use Generic Yes

Figure 2: Example Meta Data Sources

Vendor Tool Interview Process

After determining your requirements, you are ready to begin the tool vendor interview process. By conducting some research using product information that is readily available on the Internet and in industry magazines and journals, you can narrow the field of potential vendors and products to those vendors that have tools that meet your general criteria. Once you've prepared a preliminary list of vendors, you're ready to start interviewing and evaluating. Try to have the vendor come on site and use your existing meta data sources rather than evaluating on the basis of the vendor's demos.

To objectively look at all of the criteria, you should use a weighted vendor checklist to perform a tool analysis. The vendor checklist allows you to come up with a numerical score based on the criteria you specify. Figure 3 is a small excerpt of a completed checklist outlining the kinds of questions that need to be answered when evaluating meta data tools. By using an in-depth checklist, you be able to get an unbiased view of how the tool fits your company's needs as well as the tool's strengths and weaknesses.

Section/Description Weight % Met Score Comments
Technical Requirements
What programming requirements are required to support the proposed meta data repository solution (e.g., script, SQL, etc.)? 9 .75 8.4 Learning curve required.
How does the product allow multiple meta data developers to work simultaneously with the same DSS project? 6 1 6 What memory and processing requirements are needed for each user? How does the vendor suggest calculating these needs?
Meta Data Management
Is the meta data repository tool active or passive in controlling the processes of the DSS environment? If active, explain. 7 .3 21 What agents and/or triggers can be used to make the repository proactive?

Figure 3: Excerpt from Weighted Vendor Checklist

Once the checklist is complete, you can look at the scores to see just how well or how poorly a product performed. The higher the score, the better that tools fits your needs. Always remember that every company has its own unique meta data requirements and that nothing is perfect. Be prepared to compromise along the way; but if you keep your requirements and priorities clearly in mind – and manage to ignore the bells and whistles that vendors will wave in front of you – you can't go wrong.

References
1. Marco, David. "Meta Data Integration: Fitting Square Pegs in Round Holes." DM Review. September 1998. P. 26.

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