I have seen several old newsgroup items on why incremental loads in
Cognos are not worth the effort, is this really still the case? I'm
interested in hearing some other experiences in relatively similar
models (20 gig cubes with, with 5 or 6 dimensions, and only daily data
coming in).

TIA

When you say "20 gig cubes" do you mean 20 cubes each a gig in size or
several 20gb cubes?

In either case I would for sure be considering incremental builds. The 7.1
version allows you to perform time based cubes as well which may be
something worth looking at (only need to build the increment). The only
real issue to consider is that you need to ensure that you are only bringing
in new data. Rows already contained within the cube will cause the measure
values to skew.

Give us some numbers as to how many consolidated records, current full cube
build times, datasource, level of granularity you require, etc...

Glad to help just need a little more info.

ya


"Niko" <nick_wakefield (AT) hotmail (DOT) com> wrote