I have been fighting a battle to increase performance of a particular large
dimension.

I have reduced the dimension data members from over 200K to now 56K but I am
still having performance issues when adding this dimension to the pivot.

Is there a way to customize group levels? For instance,
grouping by A-D, E-G, etc, that way if the user select a group from the drop
down menu it will not take so long. Then possibly choosing what group to add
to the pivot table to reduce the time it takes to query.

I am running out of ideas to increase the performance of this particular
dimension table. Any ideas?

the problem is related to the client application
you display too many information.

if you use an application which support a paging feature, you will be able
to improve the display time.
for example, try the cellsetgrid application and see the difference.

another option is to not give access to this dimension, except through a
drillthrough command,.
or insure that you provide sheet (or analysis) allready setup with some
filters and train the users to starts with other dimensions and use the
biggest only as the latest dimension.

Grouping A-D E.... etc is good too
the user will drill down and display only a subset of members instead of the
complete list.
(as2000 provide an option to create this intermediate level for you
automatically, or you can create it easely by using a an SQL statement like
left(table.namecolumn, 1) )

good luck.


"SAM" <SAM (AT) discussions (DOT) microsoft.com> wrote

The performance issue is present in the cube browser as well. Does that make
a difference?

In creating the groups, I created a automatic group in the dimension table
but I did not see where you could customize the group to accomplish Grouping
A-D, E...etc.

I'll check again and test it. Thanks. I'll try the other suggestions as well.

"Jeje" wrote:

Thanks Jeje, I figure out the grouping and now I am testing to see if I see
an improvement.

As far as the paging feature, I am not sure. We are using Excel 2003 and not
sure where I am I would set or modify the cellsetgrid.

Now I have told others about using the biggest dimension, which is the
account name so far to be added last to the pivot table but sometimes that
doesn't seem to help with performance.

Not sure if I understand about restricting access except via drillthrough.
If the user wants to add this dimension to the row to see how the measures
are reflected, would that cause an issue?

"SAM" wrote: