by William Laurent
Deciding on a 'best of breed' enterprise metadata repository for your organization can be a time consuming and politically charged affair. Significant bi-directional and extensible integration of metadata from heterogeneous sources to a common environment will be a major task, with many stakeholders--all of whom may have their own visions of what a centralized and distributed repository or repository application suite should be.
Here are a few "checklist" topics that may jump-start your repository needs assessment and subsequent procurement efforts.
What are the Repository's Strengths? Things to Look for:
- Powerful versioning capabilities help manage and administrate complex configurations of data.
- Powerful extensibility.
- Large array of scanners/interfaces, many of which are bi-directional.
- Strong data stewardship capabilities--for instance: metadata groupings, impact analysis, searching, etc.
- Strong support of systems development standards.
- Can accommodate unstructured data. Places no limitations of types of metadata that can be included in repository.
- Repository client allows for the identification, shopping, and linking to a single business element across the enterprise. This is called data name rationalization, i.e. the automated identification of synonyms utilizing matching criteria, and is driven from a powerful data name glossary list.
- Sound separation of the repository platform from the database.
- A solid web-enabled meta data browser: Meta data 'publishing' can be highly customized.
- Inclusion of a search engine that provides potent meta data and document indexing capabilities.
- Cataloging of pointers to unstructured metadata present no limitation on what kind of metadata can be managed within the repository, providing true distributed meta data.
- Underlying relational nature of the repository enables reporting and analysis against the repository using 3rd party SQL compliant tools.
- Underlying file management system is proprietary and non-relational [can be good or bad] and does not have to rely on external resources to dictate the details of the internal architecture or API's, promoting greater efficiency in design and responsiveness to market conditions.
- Ease of extension and customization because of the separation of the repository engine from the database.
- Interactive GUI for quick utilization of the repository, offering interactive and updateable graphic representations of data. A bundled web-enabled application can change query and manipulate data in the actual repository in real time.
- Supports XML as the standard for meta data interchange, with the goal of progressively reducing the need for customized support of tool-specific environments. Accompanying modeling tool can make XML DTDs and custom repository information models (RMI).
- A true 3-tier/n-tier solution using browser based clients. May also allow applications to treat the repository contents as a large XML document, which can be viewed in various ways using XML stylesheets.
- Multitude of customizations possible. Repository of functions and services for C++, VB, and other 3GLs.
- Effective security model.
- Supports all standard RDBMS and operating system platforms.
What are the Repository's Limitations? Items to Consider:
- Proprietary nature of file management system means reporting and analysis against the repository cannot be accomplished using third party SQL-compliant query tools, negatively impacting the accessibility of the metadata.
- Administration complexity.
- Lack of BI tool interfaces and scanning.
- Unstructured data, such as Word documents, are not able to be physically stored in the repository. Instead a directory pointer is maintained in the repository to a physical location on the LAN.
- Limited OS support, specifically UNIX/LINUX or AS400.
- Read-only data shopping.
- Lack of vision or corporate instability of vendor.
- Rough interfaces and steep learning curve.
- No easy migration capabilities (i.e. promotion path/gateway from test to production) for meta data.
- Some degree of proprietary dispensation toward vendor specific ETL tools, making 3rd party ETL tool integration more difficult.
Buying a meta data repository may offer an attractive alternative to a homegrown one built from scratch. It is vital that IT management know what to look for when shopping for an extensible and scalable repository. The preceding checklist is merely a good starting point for the formulation of questions and issues that will help an organization determine the best repository for their needs.