cancel
Showing results for 
Search instead for 
Did you mean: 

performance issue while loading data from write optimized Dso to Infocube

srikanth_rapur2
Explorer
0 Kudos

Hi All,

We are facing performance issue while loading data from write optimized Dso to Infocube(FI related data), it is a full load. Every time before loading the data into the infocube it will delete the previous request.

It suppose to complete with in 8 hrs to execute 78 laks records but suddenely it is taking more than 28 hrs to complete the load.

" Conversion of characteristic values to sid"  step is taking more than 50 - 60 mins for each data packet, before it was taking only 5 to 8 mins in this step.

1). we reduced the data packet size from 50,000 to 30,000 and increased the no of parrel processing from 3 to 9, but no change in performance

2). we checked in RSRV for inconsistency of dimensional table and SID tables.But, everything is fine here.

3). we checked with the program SAP_INFOCUBE_DESIGN,we found one of the dimensional table size showing 111%.

4). sufficient back ground jobs are available, and also triggered the loads in the non business hours, but no improvement.

We are facing this issue after the system upgrade from 7.0 to 7.01, sp 14.

Could you please help into this issue.

Regards

rapur

Accepted Solutions (0)

Answers (6)

Answers (6)

colm_boyle
Active Participant
0 Kudos

Hi Rapur,

Please see Andras steps above and also my reply to his thread,

thanks,

Colm

abhishek_shanbhogue2
Contributor
0 Kudos

Hello Rapur,

If there is a provision to split the data into small chunks then try to split them and run in parallel to see if there is any improvement.

Ask you DBA guys to check for the info cube tables if they require reindexing/reorg/refresh stats to improve the performance

Thanks

Abhishek Shanbhogue

srikanth_rapur2
Explorer
0 Kudos

Hi All,

Sorry for the late reply !!!

Hi chie bill , we are not deleting the Dimension table content, it will delete only infocube content.


Hi Vengal,it is not possible to reorganize the design structure of the flow immediately.

Hi Andras, we will let you know update,  if any of these sap notes is really helps us to solve the issue.

Regards

Rapur

Former Member
0 Kudos

Hello RAPUR,

1.) Compress InfoCube before doing any data loads.


>>  Refer to SAP note: 590370.

2.) Apply corrections of following Notes if relevant to your support package SAPKW:

1827854 - Enqueue/Lock during parallel data loads
1753512 - P30:DTP:Deadlock:SM12:Between RSICCONT and RSBODSLOG
1749667 - P30:DTP:PC:IPAK:Deadlock enq. RSICCONT and RSMDATAST

3.) Additionally, you can use report SAP_INFOCUBE_DESIGNS to verify all
your cubes regarding uncompressed requests.
>> Refer to SAP note: 1461926.

Best regards,
Andras

colm_boyle
Active Participant
0 Kudos

I agree with Andras' steps, these should help. But the main problem here is that your dimension table is 111%, why are you storing so much data in the dimension, each dimension should be <30% of the volume of the overall cube. Read the note Andras provided here - 1461926 - and remodel your Infocube.

Also, reduce the number or records per request to lower than 10million as per the recommendations in the note 892513.

Colm

former_member186053
Active Contributor
0 Kudos

Hi Rapur,

Why don't you consider in the design change of data flow as it seems to be volume of data is high.

If possible, create one more standard DSO and put it between WDSO and Cube. Use Delta DTPs to load data from WDSO to Standard DSO and then to Cube. This will increase data load performance as well as it can handle huge of amount data flow into the Cube if you get more data in future.

As data increases day by day and you are doing a full load, it is trying to pull all the data from Active data table of WDSO into Cube by taking a SID look up every time.

Thanks,

Vengal.

Former Member
0 Kudos

Hi,

For the step "Complete Deletion of Data Target Contents", try not check the option "Also Delete All Dimension Table Contents"

This should save your time on SID lookup, if this help consider only delete dimension content when u can afford the times.

Regards

Bill