Successfully reported this slideshow.

Olap Cube Design

17,166 views

Published on

This slideshow gives you an overview of Cube Design in Business Intelligence.

Published in: Technology

Olap Cube Design

  1. 1. CUBE DESIGN BY HANNES MEYER OnLine Analytical Processing OLAP
  2. 2. Agenda   What are cubes?   Multidimensionality   Storage of multidimensional data.   Hierarchies   Operations   Demo
  3. 3. What are cubes?   Multi-dimensional representation of data
  4. 4. What are cubes (cont.)?   syn: Hypercube, multidimensional database (MDB), olap cube   Cubes can have more than three dimensions
  5. 5. Fact Tables   Contain numerical measurements of a certain business process.   E.g. $12.000 sales in NY store on 12-01-08   Additionally foreign keys to different dimension tables   E.g. further store/sales person information   Center in star schema
  6. 6. Dimension Tables   Contain attributes by which data can be grouped   e.g. city/region of store, product category   Linked to the fact table via their primary keys   Slowly changing dimensions: dimensions which change over time. Can be dealt with in 3 ways:   Overwritingold values   Add new row to table, distinguish records by versioning   Add new column (attribute) to existing row
  7. 7. Data Storage Models   relational databases (ROLAP)   Datain tables   Summaries stored in precalculated tables   multi-dimensional databases (MOLAP)   Data in multidimensional arrays   + Less disk space   + Better Performance (precalculated aggregates)   - Time to aggregate & calculate   - Updates require recalculation   Hybrid (HOLAP)
  8. 8. Hierarchies   Grouping of dimensions   e.g. country -> sales   e.g. month -> semester - region -> state -> city > quartal -> year -> store   2008   Germany   H1 2008   Southern germany   Q1 2008   BaWue   Jan 2008   Stuttgart   Store A   Feb 2008   Store B   March 2008   Q2 2008 …   Bavaria   Munich   H2 2008 …   Store A B C
  9. 9. Operations: Slice   Slicing is the process of retrieving a block of data from a cube by filtering on one dimension
  10. 10. Operations: Dice   Dicingis the process of retrieving a block of data from a cube by filtering on all dimensions
  11. 11. Operations: Drill Up/ Down   Drilling up: Presenting data at a higher level on the hierarchy e.g. Store -> Region   Drilling Down: Presenting data at a lower level on the hierarchy Region -> Store
  12. 12. Building the cube in SSAS   Preconditions   Connecting datasources   Defining views   Selecting dimensions   Define fact & dimension tables & time dimension   Select measures   Deploy & query the cube    Demo

×