Hi,

I'm working with AS 2000 (SP3).
I have a cube with 27 dimension and 4 measures. The data cames from 30
tables and have about 400,000 entries.
The data storage is designed to be saved up to 100MB.
When precessing this cube it begans normal. It will create 200 aggregations.
At the aggregation no. 70 the performance goes down. The processing step
becames very slow.
Does anybody knows how to find out whats going wrong?

Thx.
Leo

try to process the dimension, restart the service then process the cubes.
also, do a full process of the dimension.

"Leonardo Ellermann" <Leonardo.Ellermann (AT) modusconsult (DOT) de> wrote

Hi,

Recently i faced such a problem. Maybe apart from restarting MAS you
could also have a restart of SQL Services and also check the size of
the TempDB while processing the cube.

Try to run sp_updatestats on the SQL DB before processing and then do a
process and observe the performance.

Regards,
Sriram

Jéjé wrote:

Hi,

thx for the answer.

I've checked and tryed. But not really helps.
Can the processed data makes trouble? I've a table with 22.000 items. These
items have a code for item type. Only 200 items have a type code all other
items have the code 'n/a'. The dimension, let call this itemType, has three
levels (TypeCode, ItemNo and PackingCode). I was thinking if the relation
from 200 to 21.800 items can make trouble?

Regards,
Leo

"sriram" <sriram.ys (AT) gmail (DOT) com> schrieb im Newsbeitrag
news:1164778552.064681.140660 (AT) j72g2000cwa (DOT) googlegroups.com...
Hi,

Recently i faced such a problem. Maybe apart from restarting MAS you
could also have a restart of SQL Services and also check the size of
the TempDB while processing the cube.

Try to run sp_updatestats on the SQL DB before processing and then do a
process and observe the performance.

Regards,
Sriram

Jéjé wrote:

Hi,

I don't think it really does if you aer pulling all the records into
the cube(Including the 'na' items). If the cube/partition query
eliminates those records then you should be having a performance
problem.

Take the cube/ partition query and try to run it in query analyser, try
and sort things out from there.
1. you can find the amount of time the query runs.
2. you can try removing few of the join conditions and see if the
performance really gets affected because of any dimension.(In thi case
item)
3. from query execution plan check if you can improve the performance
with any Indexes/ Statistics in place.


Regards,
Sriram

Leonardo Ellermann wrote: