Thursday, October 4, 2012

OLAP Vs Relational


OLAP Analysis vs Relational Reporting


TOPIC
OLAP
RELATIONAL
Data Analysis
OLAP provides for online interaction with the data. Users can investigate relationships within the data with simple navigation tools. OLAP also provides context, relevance and visualization of the data.
Some relational reporting can provide limited interactivity. Overall not nearly as robust for analysis.
Numeric Calculations
OLAP technology is based upon calculations and aggregations. Cubes are very good at providing very complex calculations.
Relational reporting can also provide numeric calculations, but complex calculations are oftentimes difficult to implement.
Formatted Reports
OLAP client tools usually provide for very limited formatting capabilities. For the most part, OLAP should be used for interactive analysis, not formatted hard copy reports.
Relational reporting tools are very good at providing report formatting. Relational reporting is well suited to nicely formatted hard copy reporting.
Query Performance
OLAP usually provides for very fast query performance. The usual OLAP query is returned in under 4 seconds.
Relational reporting can be fast but oftentimes it is slow. (This is dependant on the underlying schema)
Textual Data
Although some OLAP tools will provide a means at viewing textual meta-data, overall OLAP is not the tool for storing or providing textual data. Example: OLAP cubes are not designed to store names, addresses, contact information, etc.
Relational Reporting is well suited to handle textual information.
Database Maintenance
OLAP may provide less maintenance since all aggregations are automatically provided within the cube.
Relational Reporting can be maintenance intensive if numerous "aggregation tables" are necessary.
Ad Hoc Query Creation
Most OLAP client tools make it very easy to create new "views" of the data. There are no complex joins to create. Most users can easily create their own analytic views.
Depends on how the underlying database is designed. Oftentimes user can become frustrated with table joins.

No comments:

Post a Comment