Multiple projects on a single BI environment

[Former Member]

Hi,

We are using for our project a BI environment that is already used by 2 projects.

For sure, I need now a datasource (re vendor survey flow) that is already in use by another project (I want to use the BC flow already used 'as is' by the project).

I replicated the DS from my source system to BI. Now, in BI, I can see the same flow (from DS to cube) for my source system and their source system.

1 - if I load master data and transactional data from my system using this flow; what would be the impacts on <u>their</u> flow? They should be impacted and values overwritten. Correct?

2 - what are the options of sharing the same DS (same object names) with different projects?

3 - Do we have any solutions of "mass renaming" during object copy??

Thanks for your insights.

chris

[Former Member]

Hi Chris,

If you by any possibility still are getting updates about this thread:

How did you go along with these thoughts. Did you actually implement the two projects in the same BW, but with different sources? What were your experiences from this?

Best Regards,

Rikard

[Former Member]

You havent mentioned if you are gonna be using multile source systems or not. The solution would depend on that parameter.

As such you can have as many projects/modules you like on BI system. And as suggested by Siggi, you need to make sure that the masterdata from each source system(if you have multiple source systems) is unique in nature. One way of doing it is to compound logical system to the IO to differentiate which value belongs to which source system.

[Former Member]

This is the practice:

1. You should extract only once from the source system and stage the data in BI.

2. Multiple projects should get data from the staging layer; be it master data or transaction data.

3. Usually there wont be an issue in sharing master data and in fact, master data should be shared. If you make any enhancements to the master data object and if you feel this will impact other projects, make a copy and send data from the staging to your master data object.

4. For transaction data, you must start from the staging ODS where the data will be extracted from the source system.

Ravi Thothadri

[Former Member]

Hi...Why you need two different datasource for the same data...use just one..

Regards

[Former Member]

> Hi...Why you need two different datasource for the

> same data...use just one..

>

> Regards

Hi - for example, 1 source system is CRM, the another one is SRM. Both are using the web survey/Evaluation vendor flow including their own datasources (e.g. 0WS_DATA_1, ...). In the BI side, the CRM flow already exists. I need to build the SRM one ... So both data will be different.

>You havent mentioned if you are gonna be using multile >source systems or not. The solution would depend on that >parameter.

>

Voodi, I am using 1 source system only in my project. The other projects are using several sources system.

Hope that clarifies - Thanks, chris

Message was edited by:

chris douglas

[Former Member]

Hi Chris,

normally you need to make sure that your master data is unique in all systems --> that means loading a vendor from system a with no. 0815 is Mr. Smith in 5th Avenue No. 5 in 99999 Somewhere is exactly the same coming from system b ... If you cannot ensure that a vendor with no. 0815 is the same you need to somehow harmonize the data or make it possible to identify them coming from different systems --> check this: https://websmp105.sap-ag.de/~form/handler?_APP=01100107900000000342&_EVENT=REDIR&_NNUM=184948&_NLANG...

For your transactional data it is a lot easier. Here you can just add 0logsys or 0soursystem as a characteristic to your cube, fill it from the update rules and you can separate the data by filtering on that object in the query or having different authorizations on that objects.

But anyway, getting the master data updated correctly is the toughest part in that project and there is not only one solution.

kind regards

Siggi

[Former Member]

The objective of BI is to reuse the same objects over multiple models, I think the best solution is to manage this with the other team. In my opinion, doubling everything is the worst that can be done (from a long-term perspective).