I'm experiencing some performance issues with a local cube. The cube file is
about 40Mb. (Not that big...)

My cube does contain a lot of calculated measures and custom members. So I
thought that this was the problem. But when creating a local cube with only
measures (nothing calculated) it's still the same.

If I have one measure on screen, the performance is acceptable. But I need
at least 3 measures to be able to work.

Can anyone tell me what I can do to speed this up?

When creating a local cube does he take the aggregations from the online cube?

regards,
Nico

Local cubes don't contain any aggregations, unfortunately, and that's
probably the reason why you have performance problems. I don't really know
any performance tuning tips for local cubes either (though I'd be interested
to hear any) so I guess you'll have to work around this requirement somehow
or see if you can reduce the amount of data you need to put into your local
cubes to make them smaller.

Sorry...

Chris


"verbani" wrote:

I made a 'small' error in my description of the problem. The local cube file
is between 2 and 4 MB. Not 40...


"verbani" wrote:

OK, 2-4Mb is quite small and I would have expected performance not to be a
problem at that scale. What queries are you running against the cube? How big
are your dimensions?

"verbani" wrote:

Chris,

I dont really run queries to the local cube.

But the problem was indeed a rater big dimension. If we only take a part of
the dimension the performance is good again.

Regards,
Nico

"Chris Webb" wrote: