I have designed a warehouse which includes a type 2 order dimension and a accumulating snapshot table to analyse the position of an order in the process which may help to identify bottlenecks and inefficiencies in the process.
I have a order date on the order dimension but I also repeat this date along with several other key dates in the accumulating snapshot. This table also holds a key to the order dimension.
I find myself being challenged on this design by colleagues who believe I shouldn't be repeating this field in both tables.
I believe the design is an optimum one for the following reasons:
1. It would be much more costly to derive order to date metrics from the dimension and transactional facts.
2. Providing a different view of the data for analytical purposes at a preaggregated level ensures cube and reporting performance isn't compromised.
3. Some fields such as product for example won't be stored on the accumulating snapshot and this is not the purpose of the table although attributes such as this can still be obtained by using the dim order key anyway.
I am not one to drive a idea of my own through and I always like to gain general consesnus - hence this post!
Do you feel my approach is justified and I missing any drawbacks?