Re: Slow access to MDSCHEMA_CUBES Problem

Thanks, Akshai!

Is there a section in Books Online (or other documentation) that would
describe in more detail how to combine the dimensions? I'm unclear on how you
can combine dimensions and still maintain the right level of granularity.
Would a combined dimension have a dimension table that contained all possible
combinations of the values of the old, separate dimensions?

If a document that describes this doesn't come to mind right away, please
don't feel the need to research it. You've already been very helpful and I'll
be experimenting and reading today, anyway.

--
Bob Hodgman


"Akshai Mirchandani [MS]" wrote:

I think the problem here is simply the number of dimensions... Each
dimension behind the scenes is also a cube and the schema rowset has to
iterate and analyze the dimension cubes before it decides to filter them out
of the result. I'm guessing that this is costing you a lot.

As per the other posting, if you can group the attributes together in larger
dimensions, you should see good improvements here.

Schema rowsets won't use more than one processor -- this is not an operation
that can be done in parallel.

Thanks,
Akshai
--
This posting is provided "AS IS" with no warranties, and confers no rights
Please do not send email directly to this alias. This alias is for newsgroup
purposes only.

"Bob H." <BobH@xxxxxxxxxxxxxxxxxxxxxxxxx> wrote in message
news:CE4DA67C-45B1-43B7-8EE2-E732E86EA6BD@xxxxxxxxxxxxxxxx
I have a cube with many dimensions (800). MDX queries are slow but not
disasterously so (6 seconds against 2000 row table).

The real problem is access to schema rowsets. The Profiler shows 20+
seconds
to return the MDSCHEMA_CUBES rowset.

I'm running SSAS 2005 Enterprise on Windows 2003 Server Enterprise X64 SP1
on a dual chip-dual core Opteron machine with 4 Gig of RAM.

Performance shows that no more than 30% of the CPU is ever used. Is there
a
way to encourage SSAS to use more CPU and would that speed things up (a
Windows or SSAS setting)?

There does not seem to be a great deal of disk i/o going on.

Can anyone suggest how I can approach this problem (other than changing
the
cube)? Is there a good intro to diagnosing and resolving performance
problems
in Windows 2003 Server (x64) or SSAS 2005?

Thanks!

--
Bob Hodgman



.

Relevant Pages

  • Re: Slow access to MDSCHEMA_CUBES Problem
    ... I think the problem here is simply the number of dimensions... ... dimension behind the scenes is also a cube and the schema rowset has to ... The real problem is access to schema rowsets. ... Windows or SSAS setting)? ...
    (microsoft.public.sqlserver.olap)
  • Re: Msoft wireless keyboards - need help please
    ... > Here is where the documentation was supposed to be: ... > Ron Shaw wrote: ... >> actually look at microsuxs wireless keyboards ... i dont see where msux gives any dimensions ...
    (microsoft.public.windowsxp.hardware)
  • Re: Accessing file extended properties in .NET
    ... the help files and forums and other documentation I can think of for ... some managed code equivalent to this functionality without ... a JPG's dimensions by simply examining the file and not because the data is ...
    (microsoft.public.dotnet.languages.csharp)
  • Re: Can we create Star Schema in SSAS 2005
    ... From each of those tables, you build the star schema, design the primary and foreign keys, generate the script and implement the model as your DW database. ... also, SSAS can CREATE the schema for you, in this case only a star schema ... You create empty dimensions and cubes (dimensions and cubes not connected to ... create a SCD. ...
    (microsoft.public.sqlserver.olap)
  • Re: Views as Fact Tables
    ... Unless you have only a few Mb of database, ... When you present dimension and fact tables to SSAS, ... Use named query on DSV only to apply some cosmetic change to data ... just to correctly populate dimensions and cubes. ...
    (microsoft.public.sqlserver.olap)