Skip to Content
0

RSRT Query Properties in BW on HANA - Option 0 (not to use HANA processing is selected)

Apr 13, 2017 at 09:53 AM

223

avatar image

Hi experts,

we have recently migrated our infocubes to hana optimized infocubes and when wanting to measure the query runtime, when selecting different Operations in BWA/HANA modes in the query settings at the RSRT, it gives almost the same query runtime with any other option, even when the option 0 is selected, which is supposed not to push down the operations in HANA.


Sure now in BW on HANA the queries runs faster, but I just would like to know why the query runtime doesnt change even when selected the option 0 at the Operations in HANA, so that I can choose the best processing modes for queries in HANA.

Thanks in advance!

Best Regards,

Regys

test.png (62.5 kB)
10 |10000 characters needed characters left characters exceeded
* Please Login or Register to Answer, Follow or Comment.

3 Answers

avatar image
Former Member Apr 13, 2017 at 10:42 AM
2

Hi;

Usually in the option you highlight I select "3 - Standard".

But you can test the different combinations of the query properties.

See

https://blogs.sap.com/2015/06/18/how-to-check-the-sap-bw-query-push-down-to-sap-hana/

Hope it helps;

Regards

Share
10 |10000 characters needed characters left characters exceeded
Regys Mene Apr 13, 2017 at 11:55 AM
0

Hi Ricardo,

thank you for the reply. I guess in my gui, it doesn't show the tabs that it usually should show when the query is processed in HANA. I have chosen the exact same options like in the link you sent to me, but still it doesnt show the HANA Calculation Engine layer and the Aggregation Layer Cluster Info. Does this mean, that the queries would not be processed at all in HANA? If yes, then how is it possible that with the same amount of data, the InfoCube in BW on HANA is at least 3-4 times faster than in BW on any DB...

Thank you very much.

Regards,

Regys

Share
10 |10000 characters needed characters left characters exceeded
avatar image
Former Member
Apr 21, 2017 at 03:32 PM
0

Hi Regys,

Could you please tell how you measure performance? have you disabled cache?

Please add screenshots with the execution statistics (Execute+debug, disable cache, display statistic data)

Kind regards,

Andrey

Show 4 Share
10 |10000 characters needed characters left characters exceeded

Hi Andrey,

thank you for the reply. The system is already running on HANA (BW 7.4 SP 16) and I don't know why it doesn't show the HANA processing tabs. I measure the query runtime performance as follows:

Thank you,

Regys

test.png (53.8 kB)
test33.png (31.9 kB)
0
Former Member

Hi Regys,

As I see on the screenshot most time is spent on "Empty" event(10 seconds). That could mean(in most cases) Network runtime to the end-user.

For more info regarding these events please see table RSDDSTATEVENTST.

In this case, you will not see an improvement of HANA processing, since most time is spent(according to the statistics) on the network from App Server to GUI.

To make a comparison, please select more data to be shown in List mode, and compare 6-th mode and 0 mode. If the sums of all the events excepts the empty events will be more on 6-th mode than on Zero, that is a reason to create incident on SAP support

Kind regards,

Andrey

1

Hi Andrey,

sorry that I reply you this late, but I just see your response... the Empty event I am not taking in consideration, but the difference in time is still not visible between the option 6 and 0 in data processing. The thing is that the queries are extremely full with exception aggregations and I was hoping that somehow, even partially, some of this exceptions aggregation would be pushed in HANA and with the option 6 would be somehow pushed in hana....

Nevertheless, still getting the same time (sometimes even longer with option 6 than 0) made me think that either the database settings for HANA processing are not set correctly or maybe because the query is that complex, it will not be pushed anyway to HANA...

Best,

Regys

0

What is showing on the Aggregation Layer tab of the screenshot you provided? That should show you which indices or databases are your bottleneck. If you see those pre-pended with a "$" then the index is being used and should be helping performance. If not then the complexity of your report may prevent that from being used as there are some formulas that prevent option 6 from really making any difference (I think things like %CT, %GT do that off of the top of my head)

0