SQLBits 2015

Designing a database to best support multidimensional OLAP

Data warehouse designers often ignore the specific needs of an OLAP database. In this session, John will outline the best ways to optimise your relational database to support your multidimensional OLAP cubes
Data warehouse designers often ignore the specific needs of an OLAP database when formulating their schema design. They consider the cube as 'just another reporting layer', with no specific needs. In a world where businesses need instant answers, this approach can be dangerous as it can seriously degrade your cube's load performance and make incremental loads nigh on impossible.

This session will outline the best way to optimise your data warehouse schema so that it complements, rather than fights, your multidimensional OLAP database, allowing your cube to be loaded efficiently and moreover facilitating fast incremental loads so that new data is brought on-line quickly. We will cover the design of database tables and views with specific emphasis on data types. The use of table and cube partitioning. Also, how to support incremental loads of not only fact data into partitions, but also dimensional data.