High-performance data warehousing applications begin with a consistent and unified data model – a design for performance. Design is on the critical path of high performance data warehousing. Different approaches to designing a data model for a data warehouse have frequently translated into religious wars between the followers of the star schema school and those of the operational data store (ODS) approach. However, in fact, both approaches are valid within context and represent different perspectives on the same domain or problem space. Here is how to reconcile the two different approaches.

One reason the data warehouse was invented was to avoid performance problems when decision-support queries were executed against operational systems. The long-running summaries characteristic of decision support presented logjams to the quick updates needed by the transactional systems. The situation also applies in reverse – a highly normalized data model design is not optimal for business intelligence that aggregates data across multiple transactional systems, the latter used to run the business on a day-to-day basis. Data warehousing application performance requirements are different in interesting ways from those of classical online transaction processing (OLTP) databases. OLTP applications such as SAP R/3, PeopleSoft Human Resources, Oracle Financial or Siebel operational customer relationship management (CRM) require frequent updates to what are usually highly normalized data structures. In contrast, the requirements for decision support include aggregating and summarizing of large amounts of data in order to get a business intelligence perspective on trends for customer segments and product brands, and are performed across large time horizons. For example, knowing the lifetime value of a customer requires aggregating a lifetime of detailed transactional data. Call it what you will, that mass of detail – often collected from multiple operational systems – is a data warehousing function. These differences translate into different design approaches.

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