Managing Metadata in Data Warehouses: Pitfalls and Possibilities

G. Shankaranarayanan, Adir Even
2004 Communications of the Association for Information Systems  
This paper motivates a comprehensive academic study of metadata and the roles that metadata plays in organizational information systems. While the benefits of metadata and challenges in implementing metadata solutions are widely addressed in practitioner publications, explicit discussion of metadata in academic literature is rare. Metadata, when discussed, is perceived primarily as a technology solution. Integrated management of metadata and its business value are not well addressed. This paper
more » ... discusses both the benefits offered by and the challenges associated with integrating metadata. It also describes solutions for addressing some of these challenges. The inherent complexity of an integrated metadata repository is demonstrated by reviewing the metadata functionality required in a data warehouse: a decision support environment where its importance is acknowledged. Comparing this required functionality with metadata management functionalities offered by data warehousing software products identifies crucial gaps. Based on these analyses, topics for further research on metadata are proposed. warehouse demands sophisticated back-end processing. On the end-user side, the data serves multiple decision-making needs and must be secure, of high quality, and accessed speedily. Managing metadata is essential for managing and maintaining the data warehouse. Metadata addresses many aspects of the data warehouse functionality such as data dictionary, process mapping, and security administration. Leading data warehousing software vendors now offer metadata solutions embedded within their products and other vendors offer dedicated software packages for metadata management. Together with the growing interest in metadata, frustration is growing with attempts to implement metadata solutions. Challenges are not only technical issues, but also cultural and financial ones. The software market does not address metadata needs completely: • no one product is sufficient to meet all the requirements for managing metadata, • the lack of accepted metadata standards makes it difficult to integrate metadata across products, and 2 In practice, other variants of architecture and terminology exist.
doi:10.17705/1cais.01413 fatcat:ws2bovqztrehzh3zhw44c2nuay