|
Summary tables in a data warehouse provide one with the ability to build information stores that may be aggregated data, or a collection of data that would be too complex for end users to do on a regular basis, or specific to providing performance for reporting.
So at the simplest level, I could tell you that summary tables should be used everywhere.
Summaries are driven based on user requirements and performance considerations. To address user requirements, we will build summaries that roll up data from atomic data to daily, monthly or annual
levels. Exactly how the rollups are defined will be based on needs. Other summaries are
done for
performance reasons. We had a case recently that a Cognos Cube was taking 20 hours to
build. Of
course this was not acceptable. So we used a summary table that simplified the joins and
was built as
data was being loaded into the database, so the added processing time was 15 seconds per
day. The
biggest benefit was to the cube generation process, which was lowered to 10 minutes.
Today's technology allows summaries to be built and maintained by the database with little
intervention
from users, while provided significant power. I would suggest you look at areas where
summaries will
help your and your users and that would be where I would start looking at implementing
some sort of
summary strategy.
For More Information
|
