Hi,

I have to do a dynamic distinct count in an AS2005 cube.
I have 2 formulas:

count(exists(descendants([Stores].[Store Code].currentmember,,leaves),
{measures.[#HighSales]}, "HighSales for DCount"))



count(nonempty(descendants([Stores].[Store Code].currentmember,,leaves),
[Measures].[#HighSales]))

#HighSales is a simple count measure and I have a dedicated measure group
"HighSales for DCount"

which formula perform better? :-)

its difficult to see a difference because I work on a subset of data (2000
members, 200 000 rows in the fact table while the production will have
millions and 50 000 members in the dimension) and the server has some cache.

so, from a theoritical point of view, what are your recommandations?



thanks.



Jerome.

Hi,

I would recommend you to test all your queries in OVER production
conditions.
Only this allows yout to predict behaviour all (almost all) of your queries.

Vladimir Chtepa


"Jéjé" <willgart_A_ (AT) hotmail_A_ (DOT) com> schrieb im Newsbeitrag
news:OgA7DxeHGHA.1288 (AT) TK2MSFTNGP09 (DOT) phx.gbl...