Re: CPU and MDX

Tech Tip: Click here to run a free scan for Windows Errors and optimize PC performance

have you reaggregated the cube?
specially if you have changed the cube structure after an initial aggregation.
and try do a usage based optimization.

the query is simple, returns too many rows, but simple.
a good aggregation design will certainly improve the response time.
I think that the non empty for the columns is not required, having at least 1 sales amount for 1 customer will also result in a margin value. so you'll have the 2 columns filled or none of them, its why the non empty is not required.

does the margin is a calculated member? if yes, post the code please.

also try to preload the cache by executing the query a first time.

"MIB" <MIB@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message news:8F7887AC-6974-4610-92CF-AE9AFE7DCDAF@xxxxxxxxxxxxxxxx
My query is :
SELECT
NON EMPTY { [Measures].[MARGIN], [Measures].[SALES AMOUNT] } ON COLUMNS,
NON EMPTY
{ ([DIM_CUSTOMER].[HIERAR_CUSTOMER].[Customer].MEMBERS *
[TIME].[TIME].[WeekGroup].MEMBERS ) } ON ROWS
FROM
[SALES]
and i have
for Customer only 30 000 members, for weekgroup only 36 member and my fact
table has 500 000 lines. We are in 64 bit with 4 processors dual core !!!

"Jeje" wrote:

most of the time a query will use 1 CPU only.
multi-cpu are used when you hit partitioned cubes.
(1 cpu by partition)

but before this, can you describe what you do in your query?

do you use calculated measures?
can you post formula? (and the MDX himself)


"MIB" <MIB@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:6E09F24B-E507-42B0-9171-9959FA86235F@xxxxxxxxxxxxxxxx
> We have a very long MDX request for one report. I tried every thing to > run
> it
> as fact as possible (flexible vs rigid, i have create my own aggregate,
> create cache,...). But at the end it's not going really faster, and i > am
> not
> be able to use all my CPU. Each time a queury is running, only one CPU > is
> used


.

Relevant Pages

  • RE: Troubleshooting Execution Location
    ... It's much faster on the server. ... If I run the query once via HTTP, then I change the date range and run the query again, it's almost instantaneous the second time. ... have you made sure that each Distinct Count measure is in a cube ... the axis that I think would benefit the most from aggregation doesn't show up. ...
    (microsoft.public.sqlserver.olap)
  • RE: Troubleshooting Execution Location
    ... "Jonathan Levine" wrote: ... > If I run the query once via HTTP, then I change the date range and run the query again, it's almost instantaneous the second time. ... have you made sure that each Distinct Count measure is in a cube ... the axis that I think would benefit the most from aggregation doesn't show up. ...
    (microsoft.public.sqlserver.olap)
  • help! with averaging within a cube
    ... I have a cube defined as ... My measurecome from the following query: ... data GROUP BY datepartAS datetime))) ... allow aggregation by sum and not average. ...
    (microsoft.public.sqlserver.olap)
  • Calculated Measures and aggregation
    ... My problem is that when my query just contains A, the aggregation at ... different levels of all my hierarchies gets messed up. ... SOLVE_ORDER which I explicitly set in the Cube for every measure. ...
    (microsoft.public.data.xmlanalysis)
  • Re: AS 2000 Processor busy with no query or processing activity
    ... the other place to find this option is to edit the cube himself. ... a lazy aggregation option will evaluate the aggregations of a partition ... but during this background aggregation the server is slower. ... According to Performance Monitor "Analysis Server:Connection" there ...
    (microsoft.public.sqlserver.olap)