future data size assumption of an existing SAP BW on HANA system

[Former Member]

Hello,

We have a SAP BW on HANA system (version 7.3) and currently contain a part of corporate data. A decision has been taken to enhance this system and contain whole corporate data. But the sizing of this system was not calculated to capture whole data.

We need to check existing system sizing and calculate the possible sizing growth in the future by considering the data in the SAP source systems which have not been extracted yet.

Program /SDF/HANA_BW_SIZING is strongly recomended to calculate the sizing for a BW system which is to be migrated to HANA. BUt in our case our system is already on HANA.

Can i use this program for an existing BW on HANA system to calculate the possible sizing growth (existing BW on HANA + source ERP data, not yet extracted)? or is there any other way?

Thanks in advance.

Halil

[Former Member]

Hi Halil,

Yes, you can use this report for estimation of HANA size, in case you know the amount or percentage of data to be added to BW system you can use checkbox Consider Growth in this report.

However, it is not a 100% true estimation, only the test load could show you the actual HANA size you need.

Kind regards,

Andrey

[john_hawk]

100 GB * 10% = 10 GB.

100 GB * 0.10 = 10 GB.

[john_hawk]

Hi Halil,

The general rule of thumb has been to assume a 10% compression ration, then double it.

(Memory must be available for storing AND processing)

Consider loading the data in small bites and getting some real compression ratios.

If the "part" of the corporate data is recent data (ex. last two years) and the historic data is in the same structure in the source, making a direct comparison will give you more accurate estimation than a program.

Good luck, and let us know your final solution.

John Hawk

[Former Member]

Hello John,

Thanks for your answer. Please let me give more details.

At the moment, the data which has not been extracted from source system is not history data and is belong to different processes so i can not compare with the data volume existing in BW system.

How should i consider compression ratio 10%? is the formula like expected CPU size = data volume * 0,1 or = data volume * 0,9?

Regards,

Halil