BW4 Data-Tiering-Optimization

[markus_wagner123]

Hello everyone,

Since our BW4 greenfield is constantly growing, we are currently dealing with the DTO question.

An IQ-DB for the cold store has already been set up and the first tests are promising/successful.

By using it, we primarily hope to save HOT memory at the corporate memory level.

As soon as the data has been booked into the subsequent layers, the data stock of the corporate memory could be completely pushed into the cold,

since the reporting only takes place on the subsequent layers. Thus, in the first step, there would be no loss of performance for the end user.

My question would be, is anyone already using DTO in the productive BW4 context?

How did you do it? Lessons Learned? What does the strategy / modeling look like?

How are stragglers (source system changes to data that are already in the cold) dealt with?

I would appreciate feedback!

Greetings

Markus

[RolandKramer]

Hello markus.wagner123

There are many Customers using DTO productive in SAP BW/4 environments.

Please Note, that NLS is still possible also with SAP BW/4 which is the underlying procedure for DTO.

Some Tipps for the usage of DTO/NLS:

enable SAP IQ for DBACOCKPIT (to monitor the SAP IQ DB properly)

demystifying NLS/ILM error messages … (the first defence when somethings goes not as expected ... ;-))

Best Regards Roland

[markus_wagner123]

Hi roland.kramer,

i have read almost every link which could be found on google and also a lot of content of you - thank you for the response!

As written, we were able to complete the installation and configuration correctly. changing the temperatures also works so far. Unfortunately, it gets difficult in the details.

1) i could not find any documents with information about "Absolutely do's and don'ts". I mean, what should i avoid when implementing DTO in a BW4-context and what should definetly be considered .

2) or how does DTO and a LSA++ stucture work in detail.
e.g. given Situation: I have a aDSO Z1DTO, which is filled with ECC data. This aDSO has active HOT and COLD (IQ) store, and mostly everthing is in COLD. Let's say this would be our Corporate Memory. Now we get some stragglers to this DSO for e.g. 2022.
How are these datasets processed in the next layer Z2DTO? As per this link, the changes should be loaded via changelog to the next layer. is this right?

3) or how can we do a full reaload of data in the bw layers? e.g. the Z2DTO has to be fully reloaded, but the Z1DTO has all his data already in COLD-store? Must the data of Z1DTO be switched back to HOT, Z2DTO reloaded and after that switch back to COLD in Z1DTO?

Thank you in advaced - Markus!