How do I do 'Selective Deletion' for the Cube and ODS data in the process chain? I'm working on Archive process and after moving last year's data into Archive cube/ods, I need to delete data from the main cube/ods for the last year only (not entire contents of the cube/ods)..
Is there a way to do that in the process chain?
Thank you for your help!
Have you used this method? I'm having a difficulty when creating a variant attribute for 0CALYEAR, the only option I have is 'T:Table variable from TVARVC' and the paper suggests that I should have an option 'D:Dynamic Date Calculation'. I don't have option 'D' in my system. Did you see this before?
Remember this is for 3.r
TVARC is very different in version 7 and is not used in quite the same way
To be quite honest - the approach suggested in the paper is ok for specific applications but for an archiving application like you have suggested I wouldnt use it
The prime reason for not using it is that in a BWA environment a selective deletion will trigger a full refresh of the BWA inde for the infoprovider
Th reason for this is forward compatability when the next release of BWA will not go cube to BWA but source to BWA and/or cube in parallel
Thus any change to the cube has to trigger a full refresh of BWA as there is no direct link form the cube to BWA so how does BWA know there has been a change
It is therefore more prudent to design a application that caters for this but utilising the standard load functions
I would suggest instead of deleteing you infopackage out your selections to an infosource then read them back in using a DTP and transformations - where the transformations negate the record
Compress and zero eleminate and the records will then delete on the cube in a controlled fashion
BWA will be automatically updated in delta mode
True - there will be redundant partitions BWA but as with the recommended practise for inventory cubes I would suggest each month a full refresh of the index to keep thos partition records down
Thank you very much - points are awarded.
BWA stands for Accelerator, right? We don't have it. Can you please explain further what you meant by 'utilising the standard load functions'?
If I understand correctly, your suggestion is to
1. load data I want to archive into the separate cube
2. using transformations to negate amounts load it back so amounts will become 0.
3. compress the cube .
Is that correct?
Thnk you again.
You got it - exactly that
Standard load functions - just means infopackages
I would send the data to an infosource and then to the archive cube and also from the infosource back to the source cube with a negate
This won;t work though if you have used dates as key figures..
Thank you again.
Have you ever used 'Archive' option on Extra tab on ODS maintanance? I found 'How to Archive in BW' document from SAP and they suggest to use Archive. Business Intelligence How-to Guides for SAP NetWeaver 2004 [original link is broken]
Using that option Data will be archived based on the variant into flat files and deletion will be done based on the same selection criteria and the whole process (archiving and deletion) can be scheduled as a prt of the same job. I was wondering if you ever used this function.
No I havent - and by reading the how to ... it seems a nice easy way of doing it in that scenario
Nowadays I would try and get the client to use NLS instead of the old archiving solutions - but it's a nice way of doing a one step archive and delete
(I would still prefer for my own use to use the negate infopackage just for future use of BWA)