Re: long processing time
anonymous_at_discussions.microsoft.com
Date: 02/26/04
- Next message: Steven Karpowicz: "Re: Analysis services"
- Previous message: Tom Chester: "Re: Range, Sum and a filter of other dimension"
- In reply to: Dave Wickert \(Microsoft\): "Re: long processing time"
- Messages sorted by: [ date ] [ thread ]
Date: Thu, 26 Feb 2004 08:00:11 -0800
Hi Dave,
Thanks for updating about PPU utility. I downloaded it and
currently it is processing 2 partitions in parallel. It
was easy to learn and the document along with it really
helped.
Hope this could save me some time.
Thanks
Shiva
>-----Original Message-----
>Two observations:
>
>1) Have you ran the Optimize Schema wizard in the Cube
Editor? You may be
>having long processing times because of the complex joins
being issued from
>an unoptimized schema on the Analysis Services' side. For
example if you
>have 20 dimensions, and if you look at the SQL statements
that AS issues to
>the RDBMS (SQL Server or Oracle), you will notice that it
issues a 21-way
>inner join between the fact table and *all* of the
dimension tables. If you
>"optimize the schema" using the option from the Cube
Editor, then you can
>reduce that significantly. This is discussed in both the
AS Performance
>Guide and the AS Operations Guide pointed to from:
>http://www.microsoft.com/sql/evaluation/bi/bianalysis.asp
>
>2) To process partitions in parallel, I would recommend
that you look at
>running the Parallel Processing Utility (PPU) located at:
>http://www.microsoft.com/downloads/details.aspx?
FamilyID=a2eef773-6df7-4688-8211-
e02cf13cbdb4&DisplayLang=en
>
>--
>Dave Wickert [MS]
>dwickert@online.microsoft.com
>Program Manager
>BI Practices Team
>SQL BI Product Unit (Analysis Services)
>--
>This posting is provided "AS IS" with no warranties, and
confers no rights.
>
>
><anonymous@discussions.microsoft.com> wrote in message
>news:0cd501c3fb1d$4ebfec90$a301280a@phx.gbl...
>> Hi Sean,
>>
>> That's what exactly I was looking for. Thanks for the
>> update.
>>
>> The cube which took 9 days to process has time dimension
>> as changing dimension. So I cannot divide this in to
>> partitions and need to create it from scratch.
>>
>> Document says I can use DSO to process cubes in parallel
>> through SQL server. But I'm not using SQL server at all.
>> My data comes from Oracle database. Do you recommend any
>> other DSO tool? Any steps/document on processing cubes
in
>> parallel while creating the cube would be great.
>>
>> Thanks
>> Shiva
>>
>> >-----Original Message-----
>> >Hi Shiva,
>> >
>> >You should use partitioning for this cube. It should
>> give you shorter
>> >processing times. I'd also recommend that you watch
temp
>> file creation
>> >during processing as you are more likely to encounter
>> this since you are
>> >using the disctinct count aggregate type. A lot of
this
>> material is covered
>> >in the Analysis Services Performance Guide which is
>> available on MSDN.
>> >
>> >
>> >--
>> >Sean
>> >
>> >--
>> >Sean Boon
>> >SQL Server BI Product Unit
>> >
>> >--
>> >This posting is provided "AS IS" with no warranties,
and
>> confers no rights.
>> >Use of included script samples are subject to the terms
>> specified at
>> >http://www.microsoft.com/info/cpyright.htm.
>> >
>> >
>> >
>> >
>> >
>> >"Shiva" <anonymous@discussions.microsoft.com> wrote in
>> message
>> >news:153e101c3fa55$3a9bf790$a401280a@phx.gbl...
>> >> Hi,
>> >>
>> >> I have a big cube based on a Fact table from Oracle
>> >> database. Fact table has around 270 million rows, and
>> cube
>> >> has measures with Distinct count and count.
>> >> It took about 40 hours to generate this cube based on
>> >> HOLAP but the performance was bad while accessing the
>> data
>> >> from Excel.
>> >> With MOLAP it took about 9 days, but the performance
is
>> >> great.
>> >>
>> >> Now my question is we need to process this cube every
>> >> month after adding every months worth of data.
Obviously
>> >> we can not loose 9 days to re-process this cube
whenever
>> >> we add new month's data. Is there any way to overcome
>> this
>> >> problem?
>> >>
>> >> I tried to explore creating two cubes; one is Static
>> cube
>> >> with all previous years data and one as dynamic cube
>> with
>> >> every months data till we roll in to another year.
And a
>> >> virtual cube based on both of them, but I had
troubles
>> >> merging the time dimensions.
>> >>
>> >> Any idea in this regard is appreciated.
>> >>
>> >> Thanks
>> >>
>> >
>> >
>> >.
>> >
>
>
>.
>
- Next message: Steven Karpowicz: "Re: Analysis services"
- Previous message: Tom Chester: "Re: Range, Sum and a filter of other dimension"
- In reply to: Dave Wickert \(Microsoft\): "Re: long processing time"
- Messages sorted by: [ date ] [ thread ]
Relevant Pages
|
