Converting Inventory InfoCubes, with uncompressed requests, to HANA-optimized
We faced an intriguing issue after our BW on HANA migration, wherein the Inventory InfoCubes conversion to HANA-optimized, failed.
This issue can, most likely, occur in other BW on HANA migrations and that is what prompted me to put forth this document which aims to
provide a way out for such scenarios, for my fellow BW consultants.
Context:
Converting Inventory Stnd InfoCubes to HANA-optimized InfoCubes.
Scenario:
Migration already completed to a HANA multi-node appliance.
Inventory InfoCubes still contain uncompressed, delta movement, requests.
Compression doesn’t work anymore, after migration, for multi-node (scale-out) systems.
Conversion process (RSMIGRHANADB) fails for these Inventory InfoCubes due to uncompressed requests.
InfoCube:
ZMIM_C001 – Inventory InfoCube.
The current state is that the system is already migrated to BW 7.4 on HANA SP10 and the Inventory InfoCube still contains few uncompressed requests.
These are delta movement requests. No historic movements requests.
Although the system is migrated, the InfoCube is still not converted to HANA-optimized.
We also made sure that this is an Inventory InfoCube with non-cumulative measures (Key figures):
First attempt to convert to HANA-Optimized InfoCube:
System issues an error message stating that the Inventory InfoCube must be fully compressed before the conversion:
Therefore, we attempted to compress the requests in the InfoCube:
Unfortunately, compression is no longer possible and system issues an information message that
No compression possible for Stnd InfoCubes in scale-out (multi-node) HANA systems:
Upon searching, we found the SAP Note confirming this limitation in a BW on HANA scale-out landscape
Therefore, even though the job completed successfully, it didn’t compress the request.
After perusing through multiple documents and notes, we found the panacea in the form of a tweak which needs
to be done in the RSADMIN table to force the compression to work.
Source: SAP Inventory – First Guidance doc for BW 7.4/7.5 on HANA
As per this document, we found the following important points, relevant to our scenario:
Post-migration, if you want to convert the Inventory InfoCubes without compressing the open delta movement requests* then:
Set the RSADMIN parameter FORCE_NCUM_CUBE_CONVERSION to value ‘X’
* If you have uncompressed Historic movements requests, you have to mandatorily compress them before the conversion.
Best Practice:
Post-migration, if you want to compress all the open requests and then convert the Inventory InfoCubes (which we should) then:
Set the RSADMIN parameter FORCE_NCUM_CUBE_COMPRESSION to value ‘X’
Don’t forget to:
Remove these RSADMIN parameters after the conversion/testing is done.
We followed the second (best practice) approach of first compressing the requests and then converting the
Cube to HANA-optimized.
First, we checked if the entry already exists in the RSADMIN table or not:
To insert the parameter, execute pgm: SAP_RSADMIN_MAINTAIN
Set the below parameters:
FORCE_NCUM_CUBE_COMPRESSION
value ‘X’
Confirmation
Now, we attempted to compress all the requests, one by one, and it completed successfully !
All requests successfully compressed:
Lastly, we attempted to convert the Infocube to HANA-optimized and it completed successfully !
We checked and found that the InfoCube is now HANA-Optimized:
As expected, the backend InfoCube tables structure confirms that only F table (No E table ) and Package Dimension table exist after conversion
Conclusion:
This tweak comes as a savior for the Inventory InfoCubes in scenarios depicted in this documented.
However, we should always try to compress the Inventory InfoCubes as part of pre-migration activities. This saves a lot of hassles.
Nice Share mate. Thx for sharing.
Hello Debanshu,
Thanks for the doc, very clear explanation of the issue and solution. I expect you face similar few issues post migration of Inventory Cube and you solve them and share your experience 🙂
Thanks Kiran and Syed.
Will keep sharing...and keep learning !
Reg
Deb
Wonderful document...keep sharing.
Thanks to Mukherjee.
Thanks and Regards,
Venkat