transformer auto-partitioning

[Marc Werner]

Hi,
I got a transformer model holding 5 time dimensions, 
1 big customer dimension ( 120.000 categories , 4 levels )
and another 10 small and flat dimensions with about 5-50 categories
some of them only have 1 level ( no hierarchies )

Building cubes ( we got about 50 Mio fact-records ) using that model 
having auto-partition switched on I get strange sizes of the cubes
bla.mdc
bla1.mdp  700 Mbyte
bla2.mdp  200 Mbyte
bla3.mdp  200 Mybte
...

due to transformer not being able to spread the flat dimensions across different partitions I guess.

Has anybody expierence with models of that shape ?
Is it worth to exclude the flat & small dimensions from auto-partitioning ?
Is it worth to do manual partitioning inside the customer-dimension ?

Thanks for any help
Regards,
Marc
-- 


Echte DSL-Flatrate dauerhaft für 0,- Euro*!
"Feel free" mit GMX DSL! http://www.gmx.net/de/go/dsl
-------------- next part --------------
A non-text attachment was scrubbed...
Name: marc-werner.vcf
Type: text/x-vcard
Size: 314 bytes
Desc: vcard
Url : http://lists.sowder.com/pipermail/powerh-l/attachments/20060721/582342d8/marc-werner.vcf