Query on Dimensions

Hi Guys, I am designing a star schema data for Staff Attendance process. The dimensions which I have got are like Event basically it does not have much attributes only Event Id and Event Description, Similarly I have a device dimension which contains the name of the biometric device, the location and id. My question is, is it ok to have these dimensions with so few attributes or should these be combined into a junk dimension.

Thanx for sharing the best practice.

Hi ,

if your Event and device dimension are not going to have many rows (the count of rows depends some places where we have created Junk dimension the count was around 100 rows ) then you can create a junk dimension.

thanks

The decision to junk has nothing to do with size. The question you need to ask is: does the dimension have important business meaning and is it needed for multiple facts? (Does it need to conform?)

The notion of junk originated from the problem of 'misfit' attributes coming from fact loads. After everything is mapped to conformed dimensions, you may wind up with left-over attributes you need to put somewhere. These left-overs are either discarded or placed into junk dimensions.

While there is nothing wrong with intentionally creating junk to consolidate dimensions, you need to ask if the combination created helps or hinders usability of the model.

Thank you very much for taking that extra step in explaining that thought process. Sharing this thinking is really what fosters learning. Thank you once again. Much Appreciated!