Big Cube Question - Remote Partitions

  • Hi, Brainboxes!

    We're developing a solution that is going to use a cube built off ~0.75Tb normalised data warehouse, so ending up with a cube up to 300Gb (disk size).

    As you might guess we're a little concerned about performance for both building and querying this & are looking to hardware to help us out.

    A key assumption at the moment is that we can partition our cube (along the same lines as our underlying normalised tables, ~26 partitions, ~2 years' worth of data) and then spatter the partitions about on remote partitions on separate servers.

    The idea of this is to be able to use the parallel nature of multiple servers when rebuilding our cubes (which we will have to do full rebuilds on fairly frequently as we have a few horrible dimensions which will change) as well as to be able to specify different storage models and degrees of aggregation on different partitions.

    To the point, then: has anyone had any good/bad experiences of using remote partitions?  Documentation is a little thin, and I cannot seem to find much on the newsgroups (ok, so may not be looking in the right place, if you know better please let me know!)...

    Cheers,

    Matt

  • Thanks!  Those are good links.

    I am hoping for some comments from anyone with practical experience of using remote partitions e.g. "It's great" or "It's a nightmare"!  I think we all know that there can be a gap between what Microsoft says is possible, and what is practical...

    We are going to carry out some testing with some largish volume in a few weeks, will update with findings.

Viewing 3 posts - 1 through 2 (of 2 total)

You must be logged in to reply to this topic. Login to reply