Building a Best-Fit Data Warehouse: Why Understanding Physical Database Architectures Matters

John O'Brien, CTO, Dataupia

Absract
Do you ever blame your database for not delivering the BI capabilities you expect or for failing to handle increased volumes dracefully? Perhaps it is not inherently the database’s fault. Could it be an infrastructure problem? There may be a better way to match the database and your BI workloads.

Database physical designs vary wiedely. Designs are always implemented within a set of requirements or constraints such as performance, budgets or enterprise hardware standards. Over time, physical designs may be optimized for specific applications, buyt they can stretch only so far. One physical design might perform well in an OLTP environment but fall short of the full potential of a business intelligence implementation. A data warehouse will operate most efficiently when the BI workload and logical data models are matched with the optimal phsyucial database architecture.

For the BI practitioner, this article will demystify physical database design for BI workloads such as operational BI, management BI, and analytics. For the DBA, it will describe technicques to optimizae a database for their own workloads and deliver a best-fit data warehouse.

This article was published in TDWI’s Business Intelligence Journal, January 2008. If you are a member of TDWI, you can read the complete article http://www.tdwi.org/Publications/BIJournal/display.aspx?ID=8850. If you are not a TDWI member but would still like to learn more about physical database architectures for data warehouse workloads, please contact us at info@dataupia.com.