Hub-And-Spoke: Building an EDW with SQL Server and Strategies of Implementation
SQL Server Technical Article
Writers: Mark Theissen, Eric Kraemer
Published: February 2009
To date, the implementation of a true hub-and-spoke architecture for a data warehouse environment has been an idealized and elusive goal. Although building a centralized “hub,” or enterprise data warehouse (EDW) that supports company-wide detail data is achievable, building and maintaining “spokes,” or dependent departmental data marts has proved to be the challenge. Most data warehouse environments have evolved to one of two architectures: a centralized EDW or a series of distributed and/or federated data marts.
In response to a few crucial problems with federated approaches to data warehousing, vendors and pundits alike began to promote a centralized, monolithic EDW approach. This approach has historically helped some vendors differentiate themselves from competitors and justify unreasonably high Total Cost of Ownership (TCO) for their products.
Centralization is seldom scalable or easy to manage. Conflicting end-user and management-use cases, demanding performance requirements, increasing user sophistication, and on-going demands for scalability and flexibility all work to defeat even the most efficiently managed monolithic system. In many cases, these issues cause departments and business units to create their own independent data marts.
Abstract courtesy : Microsoft
Reference : Pinal Dave (http://blog.SQLAuthority.com)