Note: this is related to a previous post relating to cube performance when using a view. The suggestion was to add a loop command to the limitmap, with the loop using a composite. However I am having trouble getting it to work.
My cube now has 8 dimensions
Currency (dense) - partitioned
Time (dense)
Model (dense)
Product (sparse)
Publisher (sparse)
Account (sparse)
Site (sparse)
Event (VERY sparse)
The cube is partitioned by currency since users must select a single currency to view their data in.
as per the origninal post, I have a performance problem when using a view against the cube in that by default it returns NA values and given that the Event dimension is very sparse it returns 1000's of rows with no value (which need filtering).
I am therefore trying to use the LOOP command in the limitmap as suggested, but cannot work out what composite I need. The cube is created in AWM and is now compressed and has global composites. However when I look at the composites that exist, rather than there being one that I can use in my LOOP command, there appears to be one per Currency (ie per partition) so I have composites such as EVT_CUBE_CURR_1_PRTCOMP, EVT_CUBE_CURR_7_PRTCOMP etc ( where 1, 7 etc are currency IDs).
Is there some other composite generated by AWM that I can use or do I need to create my own (if so how). Alternatively can I somehow combine the partitioned composites.
I tried creating my own composite using just the sparse dimensions (DEFINE EVT_ONLY_CHRIS COMPOSITE <PRODUCT PUBLISHER ACCT SITE EVENT>) and using that in the limitmap LOOP, but I get no data returned at all from my view so I must be missing something.
Thanks,
Chris