Our client machines seem to be experiencing a prolonged (<5 minutes) CPU
spike and rapid memory drain everytime an olap report is refreshed, created
or accessed. This is quite strange for me since our machines are Intel P4
based with 1GB of memory. Another weird part is that the reports are mainly
composed of only a couple of dimensions and the tables it got it's data from
has not more than 500,000 records. Any insight anyone?

Hi,

Analysis Services in some situations execute part of the "query" on the
client. E.g., if the Service Pack version is higher on the client, that part
of the query is executed on the client. If you have calculated mesaures is
executed on client etc..

Regarding solutions I would also like to have them.

One way is to enforce some connection strings on the client to force
execution on the server (when possible). Be aware that the 100% CPU usage is
passed to the server, but only executed in a CPU - not multi-treads :-((.
The "Cache Policy=7" also improves a little the execution.

"Joseph Ollero" wrote:

what is your measure?
do you use dynamic distinct count?

Try to play with your connection string.
adding something like this could help you:
cache ratio=0.001;cache ratio2=0.001;cache policy=7;non empty threshold=1


"Joseph Ollero" <jollero21 (AT) yahoo (DOT) com> wrote