Tuesday, March 27, 2012

Advice on database volume

Hi all,
I'm new to datawarehousing. I've built a DW with two fact tables, each one
will be added with more than 5 to 10 millions rows each day. Do you think
that this volume is compatible with Analysis Services 2005 or do I have to
change the model I use, by creating a fact table for each customer I have
(for instance).
Second question: if the awnser of the previous question is that I can keep
the model I've built, will I have to investigate cube partition ?
thank's.
JeromeYou can work with this volume of data in Analysis Services 2005 without
changing your schema (and I strongly suggest you to not change your
schema!).
You can use partitions to handle this volume of data without having a
single big partition that is hard to process if for whatever reasons
you need to full process your cube.
Be careful with aggregation design - you really need preaggregations on
this volume of data.
Marco Russo
http://www.sqlbi.eu
http://www.sqljunkies.com/weblog/sqlbi
Jerome wrote:
> Hi all,
> I'm new to datawarehousing. I've built a DW with two fact tables, each one
> will be added with more than 5 to 10 millions rows each day. Do you think
> that this volume is compatible with Analysis Services 2005 or do I have to
> change the model I use, by creating a fact table for each customer I have
> (for instance).
> Second question: if the awnser of the previous question is that I can keep
> the model I've built, will I have to investigate cube partition ?
> thank's.
> Jerome|||Hi Marco,
What do you mean exactly by "aggregation design"?
What is exactly preaggregation? Do you know some references on the web?
Thanks
Jerome
"Marco Russo" wrote:

> You can work with this volume of data in Analysis Services 2005 without
> changing your schema (and I strongly suggest you to not change your
> schema!).
> You can use partitions to handle this volume of data without having a
> single big partition that is hard to process if for whatever reasons
> you need to full process your cube.
> Be careful with aggregation design - you really need preaggregations on
> this volume of data.
> Marco Russo
> http://www.sqlbi.eu
> http://www.sqljunkies.com/weblog/sqlbi
>
> Jerome wrote:
>|||yeah aggregation % in Analysis Services
-Susie
Jerome wrote:[vbcol=seagreen]
> Hi Marco,
> What do you mean exactly by "aggregation design"?
> What is exactly preaggregation? Do you know some references on the web?
> Thanks
> Jerome
>
> "Marco Russo" wrote:
>|||yeah % aggregation in analysis services
Jerome wrote:[vbcol=seagreen]
> Hi Marco,
> What do you mean exactly by "aggregation design"?
> What is exactly preaggregation? Do you know some references on the web?
> Thanks
> Jerome
>
> "Marco Russo" wrote:
>sql

No comments:

Post a Comment