SSAS Cube
-
We have a relational DB in SQL Server 2008 which grinds up monthly data sets based on a set of criteria, these runs can take between 2 and 4 hours to produce and there may be 10-12 per month in 3-4 run sets. I am proposing that each run set has a cube for reporting and analysis. The results are stored in 2 (fact) tables, each table has a view which pulls in the dimension fields, date, branch, product etc. Each Run can result in +7m rows Q1 Should I replicate the data into a reporting database before building the cube or build direct from the relational DB? Q2 It has been recommended that I use the views, do I also need to pull in the dimension tables or can I use the fields in the view? eg Product dimension is serviced by a distinct of the Product column (answer - use the dimension table for cases where there is no record for a dimension element this period :doh: ).
Never underestimate the power of human stupidity RAH