BW4 Unique Data concept for non-SAP and SAP Sources

[svenhertrich]

Dear colleagues,

i am asking a very general question, which by nature has nothing to do with a release of a BW system. Nevertheless i want to add this information because may somebody has besides the old threads on my question new ideas by using the new concepts and methodologies of BW4.

Background:

We are going to migrate / move our old outdated Netweaver 7.3 to a new greenfield BW4. We had in our old system conceptual problems with uniqueness of master data. We are connecting against two SAP ERP systems (each division has separate ECC) and since couple of months also against various non-SAP ERP systems (all different companies). Our reporting solutions are supposed to work for local usage and as well for global usage (cross company, cross divisional). We always had difficulties with uniqueness of data. We created few custom info objects in order to meet this requirements. We also isolated data flows and objects from each other, which caused us double maintenance in case of changes. We never had a clear concept we just made it somehow working so everybody could run reports (local and global)

Now:

We have now an empty box BW4 where we can redesign our applications (what a great opportunity!!!) In our old system we have not used most of the standard contents which were available. We want to change this as well and go with business content from sap wherever we can do (still 80% of our reports are running on SAP data). It will for a lot of cases already fit our requirements.

But the question arrives here. We always have to deal with 2 worlds. The SAP and the non-SAP. From the SAP we can control already a lot since SAP framework (InfoObjects, Datasources etc.) is giving a lot of services. For the non-SAP we have build everything (god bless the field-based modelling) - and we want to combine for our Clients the views. For a standard sales report for instance our users should not notice if data is retrieved from SAP or non-SAP. But in terms of LSA and Masterdata we don't want to maintain everything x-times. Therefore i am looking for a smart concept how to integrate this in the BW4.

I've read a few articles related to compounded infoobjects with 0SOURCE_SYTEM - which would actually give the flexibility i believe its needed. We could maintain the same Dataflow for multiple datasource origins. As well as Query and Report definitions could run always on the same Infobjects. But most of the threads are turning out with a "try to harmonize the data in another way" since the global combounding on the entire BW4 system is giving a lot of headache.

Is there anybody out there who is facing similar questions like mine and has a good advice for me?

Many thanks in advance.

[Jan_van_Ansem]

Hi Sven,

I agree with Frédéric that compounding your master data with the source id would be a way to resolve conflicts. This is particularly helpful if you have two of the same keys (say, for customer or material) which ahve a different meaning (as they originate in different systems).
This is not helping you with unifying the data - so if you have two different keys which both represent the same object.
Take for example this dataset

System_ID Customer_ID Customer Name
S_001 1001 Fast_Cars_Inc
S_002 1001 We_sell_cars_ltd
S_002 1002 Fast_Cars_Inc

In this dataset you have two problems:
1. The same customer (Fast_Cars_Inc) has two different keys (originating from two different systems.
2. Different customers have the system Customer_ID (coming from different source systems).

The compound key in BW lets you create this dataset (as System_ID is part of the key) but doesn't help you when you want the aggregated results of 'Fast_Cars_Inc'.

To deal with this problem, you would need a different attribute, which is a Global Customer ID.
This could either be a pointer to a leading system (S_001|1001) or a new key all together (GC00001). If you create a new Global Customer object, you might have to think about how you are going to maintain master data. You could automatically load some attributes based on the master data in other systems.

System_ID Customer_ID Customer Name Global_Customer
S_001 1001 Fast_Cars_Inc GC0001
S_002 1001 We_sell_cars_ltd GC0002
S_002 1002 Fast_Cars_Inc GC0001

If you are going to consolidate master data, questions about 'who is owner of which record/attribute' will need to be addressed.

Of course, if you want a solution with all bells and whistles, then you need to think about Master Data Management / Master Data Governance (MDM/MDG). This would integrate master data from different operational systems at transactional system level (this can be SAP as well as other systems).
This would be a much larger project though, as it will have an impact on the operational processes.

[FCI]

Hi Sven,

You have mainly 2 ways to handle this kind of scenario: compounding (with the source ID) or concatenated key (with the source ID too).

There seems to be a consensus toward the concatenated key scenario (more flexible and with less constraints than the compounding one)

See here for some of the advantages:

https://books.google.fr/books?id=JAHcBgAAQBAJ&pg=PT277&lpg=PT277&dq=concatenated+key+compounded+sap+...

Regards,

Frederic

[svenhertrich]

Hi Oscar,

Thanks for your response. We are experimenting with SAP HANA Smart Data Access. This allows me to easily read the data from other sources. But my challenge is more to combine this data in unique reporting structures.

I am looking for a best practice for a global data-warehouse which allows same reporting features for non-SAP and SAP sources. I want to use same Security, Same InfoObjects, Same attributes.

[Oscar_Alberto]

Hi Sven.

For what you comment, your main problem is with the data that comes from non-SAP sources. For these, have you considered using SAP SQL Data Warehousing with HANA? This approach would give you the necessary freedom to model the data and perform the cleaning of them and use it in mixed scenarios with BW/4HANA.

You can also use SAP HANA Smart Data Access, which allows you to consume data from other databases directly in HANA with the option to extract or not the information.

I hope to be helpful.