implementing junk dimension

Hi everyone, I am new to DW and have an implementation question regarding a junk dimension.

I have a junk dimension with a 15 Y/N flags. Total number of rows for all possible combinations of these flags is around 32,000. Of those only 57 combinations are being used by the facts. So I want to just load these 57 rows in the junk dimension and use those foriegn keys for loading the fact table.

The question is when a new fact arrives with a new combination of these flags, how to create a new row in the dimension and get the key populated into the fact table for that fact using SSIS or SQL. I mean I want to generate the dimension rows as and when they are needed as opposed to having them all created ahead of time. I dont know if thats OK

Hope someone can help me with this

You load dimensions before you load fact tables. When the new dimension row comes in, insert it. When you load your fact table, the dimension row will be there.

The problem is you might not know about new combinations for the junk dimension until you start loading the fact table. One approach is to pre-scan the incremental fact rows to find new combinations and add them to the junk dimension. See Design Tip #113 for more specifics:
http://www.kimballgroup.com/html/09dt/DT113CreatingUsingMaintainingJunkDimensions.pdf
--Warren

Where is this file: DT113CreatingUsingMaintainingJunkDimensions.pdf? The link http://www.kimballgroup.com/html/09dt/DT113CreatingUsingMaintainingJunkDimensions.pdf is broken.

It's here: http://www.kimballgroup.com/2009/06/03/design-tip-113-creating-using-and-maintaining-junk-dimensions/

Tell you what, if it's only 32,000 with Y and N, I would not bother to maintain the junk dimension as it's a very small dimension for a lot of attributes. I would just prebuild by cross joining Y and N sets.