My first Publication Agile-Data-Warehouse-Design-eBook | Page 47
26
Chapter 1
Summary
Data warehouses and operational systems are fundamentally different. They have radically
different database requirements and should be modeled using very different techniques.
Dimensional modeling is the appropriate technique for designing high-performance data
warehouses because it produces simpler data models—star schemas—that are optimized for
business process measurement, query performance and understandability.
Star schemas record and describe the measureable events of business processes as fact tables and
dimensions. These are not arbitrary denormalized data structures. Instead they represent the
combination of the 7Ws (who, what, when, where, how many, why and how) that fully describe
the details of each business event. In doing so, fact tables represents verbs, while the facts
(measures) they contain and the dimensions they reference represent nouns.
Dimensional modeling’s process-orientation supports agile development by creating database
designs that can be delivered in star schema/business process increments.
Even with the right database design techniques there are numerous analysis challenges in
gathering detailed data warehousing requirements in a timely manner.
Both data-driven and reporting-driven analysis are problematic, increasingly so, with DW/BI
development becoming more proactive and taking place in parallel with agile operational
application development.
Iterative, incremental and collaborative data modeling techniques are agile alternatives to the
traditional BI data requirements gathering.
BEAM✲ is an agile data modeling method for engaging BI stakeholders in the design of their
own dimensional data warehouses.
BEAM✲ data stories use the 7Ws framework to discover, describe and document business
events dimensionally.
BEAM✲ modelers encourage collaboration by using simple modeling tools such as whiteboards
and spreadsheets to create inclusive data models.
BEAM✲ models use example data tables and alphanumeric short codes rather than ER data
abstractions and graphical notation to improve stakeholder communication. These models are
readily translated into star schemas.
BEAM✲ is an ideal tool for modelstorming a dimensional data warehouse design.