Skip to Content
avatar image
Former Member

Wrong copa data have to be delated from Cube in production

Hi All,

I have some wrong delta data uploaded in Production system in last month. Now the customer’s business process has got changed and according to that I have made changes in update rules.

Now I want to correct the uploaded data in production system. There are around 7000 records wrongly uploaded on 31st of July. Before this month end I want to correct these records as the new data would be uploaded by this month end with updated business process.

Could any one please tell me how to do the correction I am little bit scared as I have to it in production system..

The scenario is like this.

1. Copa data comes in ODS.

2. And delta data moves from ODS to cube

The logic which I have though is:

1. I will do the selective deletion in cube for the wrong data.

2. and will do the delta upload from ODS. (But I am confused how to this)

The suggestions from all experts will be highly appreciated with reward points.

Thanks,

BW USER

Add comment
10|10000 characters needed characters exceeded

  • Get RSS Feed

3 Answers

  • Best Answer
    avatar image
    Former Member
    Aug 21, 2007 at 09:39 AM

    Hi,

    the first thing is to identify the scope of these 7'000 records.

    Do they belong to one single fiscal period? To the same company code?

    Try to find a characteristic for which you don't have to much data in your target cube.

    Let's assume that the 7'000 records are all belonging to FISCPER 2007.007.

    Check in your cube and in your ODS how many records you have with FISCPER = 2007.007.

    If the amount is reasonable just selectively delete your data in your target ICube and then perform a full load from your ODS to your cube with a selection FISCPER = 2007.007 (ensure that no data is loaded in your ODS; that all request are activated in your ODS and that your new URules are active!)

    This won't impact you datamart deltas; after this procedure, ensure that your data is accurate and restart your loading scenario as usual.

    hope this helps....

    let us know how it goes...

    Olivier.

    Add comment
    10|10000 characters needed characters exceeded

  • avatar image
    Former Member
    Aug 21, 2007 at 05:15 AM

    Hi, BW user ,

    U can do the selective deletion from the infocube. But loading the delta into Cube is a bit risky ,because cube has the additive capabilty and any addition may incorrect the data.. If the data amount is not big do the full load again from ODS by deleting the previous load...

    Or else delete the data from cube ..

    De;ete that particular delta from ODS , and reload it from PSA and reload the delta into cube .. It will work because any changes wil the delta for a cube..

    Hope it helps..

    Add comment
    10|10000 characters needed characters exceeded

    • Former Member Former Member

      Hi,

      Do the selective deletion data for last month in the cube and make a full repair request from ods to cube for the last month .(in infopackage you can make selection for the calmonth)..

  • avatar image
    Former Member
    Aug 21, 2007 at 09:27 AM

    Hi,

    Since the requests in the ODS would have been datamarted, this is going to be a tricky issue to correct. If possible, you can resolve this issue with reverse loading:

    1. Create 2 temporary cubes as a copy of the original COPA cube. Generate export datasource on the original cube.

    2. Create update rules from the original COPA cube to the temporary cubes with automatically proposed mapping. Add logic to reverse the key figures in the start routine for load into temp_cube1 and the new logic (as per business change) into start routine for load into temp_cube2.

    3. Selectively load the temp cubes from the original cube so that the wrongly uploaded records are now loaded into the temp_cube1 with reverse key figures and into temp_cube2 with the correct logic.

    4. Generate export datasource on the temp cubes and create update rules for loading original cube from the temp cubes with automatically proposed mapping.

    5. Reload the reversed data into the original cube without transformation.

    6. Reload the correct data into the original cube without transformation.

    I hope this would have helped.

    Cheers,

    Reetu

    Add comment
    10|10000 characters needed characters exceeded