"Matt Beaverton" <nothankyou@nowhere.edu> wrote in message
news:Xns94FEB59B16AB8j7G5qVz0xU2@207.46.248.16...
> Hi Eric,
>
> I guess you understood I wanted to break "1 year cube" into "12 month
> cubes"? No, I didn't want to do change anything at the cubes side.
>
> Each cube is currently processed from a big fact table. The fact table
> is really very big. And I'm considering to split the fact tables into
> smaller physical tables. Next, create a partitioned view to UNION all
> these small tables to make it look like the former big fact table.
>
> Therefore the cubes still see the same fact table.
>
> I'm not familiar with Analysis Service. I'd like to know if it works OK
> with a partitioned view. I expect it does and if so, is there any
> advantage, shortcoming, design change?
>
> Thanks,
> Matt
>
>
> "Eric.Li" <anonymous@microsoftnews.org> wrote in
> news:usA2MdnSEHA.2112@TK2MSFTNGP11.phx.gbl:
>
> > Well it will help the performance because each query will scan a
> > smaller cube unless user requests to return everything. It won't help
> > your processing time much because you still have same amount of data,
> > just split it into small chuck. Actually it may hurt it as it needs to
> > access the view multiple times.
> >
> > Your DW will become more complicated because now you have multiple
> > cubes, each one with its own dimensions. If users want to access more
> > than 1 months data, they will have to modify their current reports or
> > you have to create virtual cubes for them.
> >
> > One more thing, if you make changes in one cube, you have to duplicate
> > it across all other cubes and we know how user friend that cube editor
> > is. Especially if you have alot calculated measures, it can be really
> > painful.
> >
> > Also, depends on your design, you may new cubes for each new month, or
> > may be not, all depends on how you cut your fact table.
> >
>