#BWonHANA: Standard DSO also Optimized for HANA
The activation operation of a standard data store object (DSO) in BW 7.30 on HANA has been optimized for HANA. It has now all the performance advantages of the HANA-optimized DSO which has been originally shipped and that required an explicit conversion step during the migration to BW-on-HANA. This is now obsolete. Also, older data flow setups can now benefit from some of the advantages. The necessary changes are planned to ship with BW 7.30 SP10 (7.31 SP9 resp.) and HANA revision 57. Alternatively, BW 7.30 installations on SP8 or SP9 (7.31 SP7 and SP8 resp.) can apply code changes offered in OSS note 1849497. In summary, this yields a significant simplification.
The availability of the active and not-active data concept in HANA allows to maintain a materialized change log of the standard DSO at no impact on main memory consumption. It is even more efficient as the (on the fly) calculated change log of the HANA-optimized DSO consumes quite some memory at the moment it is accessed. So overall, the reworked standard DSO provides all the advantages and requires no conversions. As a consequence, the HANA-optimized DSO becomes obsolete. It will still be supported. So don’t worry if you are using this type of DSO. It continues to work as before. A re-conversion tool will be provided in June. Details can be found in this document.
This blog is cross-published here. You can follow me on Twitter under @tfxz.
Excellent!! Thanks Thomas for sharing this update.
So now onward who ever moves to BW on HANA from BW on RDBMS does not need HANA Optimized DSO?
Or there will be some scenario where using HANA Optimized DSO would be beneficial?
yes, the former HANA-optimized DSO is obsolete. Don't convert existing (standard) DSOs anymore!
OSS Note 1849497 is not yet released for general use. Any idea when this will be released?
we'll release the note with availablity of this feature for SP8/SP9: this is planned for end of May.
Hi Thomas / Rainer,
This is great news for giving all DSO's the speed of HANA.
I am assuming there will also be a Advanced correction for 7.31 SPS6/SPS7 and ultimately be a part of 7.31 SPS8 ?
yes, the correction will be also available for 7.31. We plan for SP7+8.
The details will be mentioned in the note, which will be released as soon as we release the correction to customers.
Thanks Rainer for the prompt reply.
Is HANA rev 57 a must? It seems that rev57 is not available yet.
rev 57 is planned to be released next week.
Thanks for the information, is there any update of REV 57 availability as Dev Edition instance ??
what are the posibilities in HANA sidecar in BW context?
Is it possible to move DSOs into the sidecar HANA?
Including loading and reporting features?
Could any one confirm that, note 1849497 can be applied for BW 7.3 SP8 version to have HANA optimised DSO's. The reason is, the note says, its applicable for BW 7.3 SP10 and nothing about 7.3 SP8. All online links or SAP docs says that the note supports SP8 as well..Pls confirm
Please explain me how it can happend :
If i correctly understood,
at first in HANA you delete Change Log table and replace it by Index and Calculation View, so in fact there are no physically table and memory saved;
now in HANA is restore separate table for Change Log and in presentation we can read follow feature info: "Performance advantages for propagation into multiple InfoProviders and less resource/memory consumption due to persistent Change Log "
Please explain thereby preserved the memory because we have a new table?
And one more quastions. If we have standart DSO not optimized to in HANA DB we have all data from three table Actine, Non-Active and Log, yes?
This sort of hints at an interesting topic for me, memory management. I've read many opinions on HANA and almost all of them infer or explicitly indicate that further modeling of data is not required. Don't we still have to consider the memory workspace available for query execution? For example if all models in BW/HANA stored the data in the lowest granularity, you have to bring in more indexes into memory to calculate the end result, correct? So if you have many users pinging your HANA database, the memory workspace required for a non aggregated model will be greater than an aggregated model. Is this an accurate statement? I'm not saying we need to do cube aggregates like in old non accelerated BW days but about modeling say the MM documents from document level to metric level(shrink, receipts, transfers,etc).