views:

45

answers:

1

Hi Folks,

I'm in the middle of designing a SSAS db. I get the theory and the use of this stuff. Here's the thing, I've got a logging database that logs interesting order statuses which I would like to measure time to complete. I've got these tables (not implemented), to measure status times

time_dimension

user_dimension

status_dimension

status_fact - dimension references and timeInStatus measure

So my question is, do I create regular database and stage these things up for an SSIS task to pull into a SSAS db, or do I just create an SSAS db and describe the regular db with SSAS?

Naturally I'm new at this, but this type of analysis has been an interest of mine for a looong time! Your help is appreciated.

+1  A: 

If your source DB (the logging one) is really nicely normalized around the data you need, you can probably get away without the stage.

Performance may suffer, development may suffer, etc.. I think a DW (stage) db is almost a necessity to fully leverage SSAS though...

Shlomo
fair enough, I'm looking for the design guidance. This does the trick! One of my measures I plan on calculating into a staging db anyhow. I don't know how else I can get it. The measure is a diff between a date stamp on this record and its previous. I wasn't sure if a "snowflaked" style db was necessary. Thank a bunch.
Josh Robinson
Just to add on, the stage db may be a bit more work up front, but I do think it reduces maintenance as well...
Shlomo