Data Marts, Conformed dimensions and Data Warehouse

[Sophie]

Hello,

Can you please assist me in understanding the following. I am new to data warehouse. I'm using microsoft products to build my enterprise DW/BI

1. Do I build data marts for my enterprise DW/BI or do I iteratively build a data warehouse? I currently have a Sales Cube which feeds from a database(datamart). Do I build another database using the same dimensions (duplicating the data )for my Inventory cube?

2. I'm pretty much in the gray area. I am using the Kimball methodology which suggests building data marts and using conformed dimension, but I'm not sure if this means reusing the dimensions that are in my cubes or replicating the data accross data marts?

3. If I want to define friendly names for my dimension members do I do this in SSAS.

4. What's the physical design of a data mart? Can it essentially sit in 1 database ?

Thanks for you help.

Regards,

Sophie

[ngalemmo]

1. You iterativly build a data warehouse. Fact tables in the data warehouse should be at the lowest grain possible. Facts share the same dimension tables. These become the source for aggregates (i.e. MDX cubes). The data warehouse, generally, is a single database. The MDX cubes can go wherever appropriate.

2. You reuse the dimension tables. What winds up in the MDX cubes are attributes of those dimensions which are usually at a summary level.

3. Don't know... I assume you do it in the data model.

4. The relational DW (staging, fact and dimension tables) should coexist in the same database (there are other architectures, but as a beginner, let's keep it simple).

[Sophie]

Thanks. This was helpful.

[Sponsored content]