core data warehouse-component in SAP BW

[Former Member]

Hi experts,

a typical data warehouse reference architecture (as described in many books) basically consists of a core data warehouse, an ods and data marts.

As far as I understood in SAP BW an ODS object acts as a storage location for consolidated and cleaned-up transaction data (transaction data or master data, for example) on the document (atomic) level. This data is not stored permanently but overritten as soon as changes are applied to data in the operational systems.

But where in SAP BW is a component like a core data warehouse that stores consolidated and cleaned-up transaction data on atomic level permanently without overwriting it????

Thanx

Axel

[Former Member · Accepted solution]

Hi Axel,

I guess pizzaman was refering to this PDF:

https://www.sdn.sap.com/irj/servlet/prt/portal/prtroot/docs/library/uuid/5586d290-0201-0010-b19e-a8b...

You can refer from page 17 onwards.

Bye

Dinesh

[Former Member · Accepted solution]

hi Dinesh

thanx for the PDF! Does it refer to SAP BW 3.5 or NW2004?

Do I get it right that in SAP BW version 3.5 there is no "core data warehouse" that provides persistent historic data on a granular level but only ODS-Objects that can only provide the latest integrated but NOT HISTORIC data on a granular level?

And do they actually introduce a new component in NW2004 (referring to page 20 in PDF) or only suggest a differnt way of modelling a set of ODS-Objects and InfoCubes (that have already existed in version 3.5) with granular data to a "Data Warehouse Layer"?

Thanx

Axel

22 von 38

[Former Member · Accepted solution]

hello all,

I 've come to the conclusion that Pizzaman is right. The "core data warehouse" or "Data Warehouse Layer" (in SAP terms) consists of different levels of ODS-Objects (with the second ODS-layer holding persistent granular data) and Master Data as shown on page 22 of PDF.

Thanx to Pizzaman

Thanx to Dinesh

Axel

[Former Member]

Might want to review the Modleing the Data Warehouse Layer with BI article on this page

[original link is broken]

As to slide 20, (BW 3.5 = NW 2004, BI 7.0 = NW 2004s) , you can implement the this approach in 3.5, there's nothing new in NW 2004s that provides this, it is a desing approach. 2004s does change the terminology a bit, and has reworked the ETL side of things and added the Transational Data Store, all of which helps implement this new layered approach, but I don't think think the delivered Business Content has really caught up with the idea - perhaps with newer modules.

[Former Member]

That would be a Cube. You cannot overwrite any data in the cube. ODS is usually used to check if the data that is being collected is the correct data and free of any errors. Then it is loaded into a Cube. You can run queries on that cube and get the required data.

Hope that answers your question.

[Former Member]

I don't think a Cube would be the core data warehouse, storing the atomic data. A cube stores aggregated data that typically has gone thru various transformations as well.

Think there's a paper somewhere on SDN about BW layers, or maybe it was in SAPInfo. Some of the newer modeling recommendations have been to use the ODS structure to provide both the atomic opreational data sore - unaltered data as received fom the source system, and then send it to another ODS (now just called DataStore in 7.0 ( which is now called NW 2004s)) where you would perform transformations, updates, etc.

So the ODS can provide both - it just becomes a modeling choice. If I can find the article tomorrow, I'll post a link.

[Former Member]

hi pizzaman,

that would be nice!

Axel