We have a very large dimension, nearly two million members on the lowest
level. On average, only 75k actually have facts for them in a partition.

In setting member counts, would the proper setting be two million or 75k?
The aggregation wizard uses this number and I want to make sure it's set
right.

In a nutshell you should use the number of members per partition, but
this does depend on your partitioning scheme. The "Microsoft SQL Server
2000 Analysis Services Performance Guide" has details on this at:

http://www.microsoft.com/technet/pro...ntain/ansvcspg.
mspx

There is a section specifically on this titled "Decrease Dimension
Counts in a Partitioned Cube". If you are dealing with a dimension of
this size I would encourage you to read the whole article.

HTH

--
Regards
Darren Gosbell [MCSD]
Blog: http://www.geekswithblogs.net/darrengosbell

In article <uLImw566FHA.808 (AT) TK2MSFTNGP09 (DOT) phx.gbl>, jesperzz (AT) hotmail (DOT) com
says...