Mapping for "Junk dimensions"
490792Mar 23 2006 — edited Mar 23 2006I'm trying to implement a mapping for some columns in what Kimball calls a "Junk Dimension." There are 25 columns in the dimension with cardinality's ranging from 2 to 10. There are 100,000,000 possible values, but only around 100,000 actually exist in the data set. I'd rather just populate the table with the 100,000 rows.
Here's a refernece to Kimball's discussion of this:
http://www.kimballgroup.com/html/designtipsPDF/DesignTips2003/KimballDT48DeClutter.pdf
Is the only way to do this in OWB to use a deduplicator (DISTINCT)?
If I have 2 dimensions like this, plus the fact table, am I going to fully scan the source table 3 times, once for each of the 2 distinct queries, and once for mapping into the fact table?
Is there a more efficient solution?