Saturday, February 25, 2012

cube partitions strategy

Hi,
I have a product dimension with the following hierarchy:
Category->Brand->Type->...->UPC
There are 10 category items in the hierarchy. In every update, we will add some new products under 1 or 2 categories, which will cause rebuilding the whole cube, which takes very long time.
Can I partition the cube along the category level so that only the affected partition needs to be rebuild in each update?
or what's better partition strategy here?
Thanks for your attention.
That's probably what I would do, cut the partition the along categories
so only affected category will be reprocessed, then create a virtual
cube on top so from users' point of view, there's one cube.
Or you can incrementally update your dimension but it's more complicated
because you have to be careful about the members and calculated members
to avoid double count.
Eric
yongliyang wrote:

> Hi,
> I have a product dimension with the following hierarchy:
> Category->Brand->Type->...->UPC
> There are 10 category items in the hierarchy. In every update, we will add some new products under 1 or 2 categories, which will cause rebuilding the whole cube, which takes very long time.
> Can I partition the cube along the category level so that only the affected partition needs to be rebuild in each update?
> or what's better partition strategy here?
> Thanks for your attention.
Eric Li
SQL DBA
MCDBA

No comments:

Post a Comment