Hello everybody,
I’m trying to find a way to reduce the total time to create a time-base partioned cube.
Anyone can suggest a solution about the following crazy idea:
If i have a model that builts a time-based partioned cube on 12 months, as you know, I’ll find at the end of the job 14 files (12 childs cubes, the main cube and 1 vcd) .
This job takes a lot of time and I’d like to reduce it.
I’would like to create 12 cubes (one for any month) by 12 parallel processes and to find a way to built the 13th main cube by a separate process (about the vcd there’s no problem).
If I found the way I could say “Bingoo” because I can reduce a job from 24 hours in 2 hours.
won’t work, transformer is a single threaded application, so the cubes are build sequential. :’(
you should look at other things like:
prepare aggreated data, which transformer reads -> is quicker
if the database is oracle there are some fetch rows options to read the data much quicker
[quote=“CognosGuru, post:2, topic:536”]won’t work, transformer is a single threaded application, so the cubes are build sequential. :’(
you should look at other things like:
prepare aggreated data, which transformer reads -> is quicker
if the database is oracle there are some fetch rows options to read the data much quicker[/quote]