SSAS Cube process

Question

SSAS Cube process

maheshsahini

SSC Journeyman

Points: 94
More actions
July 21, 2011 at 2:21 am

#244687

Hello friends,
I work for a huge retail chain and so there is large amounts of data that pump-in daily, recently we have decided to go for a cube to help us analyse the data better.
the cube we have takes almost 4 hrs to get processed (full), But from what i read from the forums I understand that in full process entire data is dropped and then pumped-in again.
I also tried the incremental update option while processing the cube, but its only for measure groups.
What is the optimum way to process both dimensions and measure groups, without having to drop the entire data in it (as in case of fullProcess)
All I want to do is come down with the processing time of the cube.
Any help would be appreciated. Thanks in Advance.
Regards,
Mahesh

Viewing 4 posts - 1 through 3 (of 3 total)

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

VALEK SSCarpal Tunnel Points: 4206 More actions · Answer 1

1. Read more about the modes of processing and how they work. This bit is usually overlooked by the BI developers and isn't understood very well.

Everyone uses FULL and thinks that this is the simplest and reliable way of populating the cubes.

2. Measure how long it takes to process individual items (dims/measure groups).

Thy to optimise processing of the dimensions first. Customers and products sometimes as massive and can be processed differently/separately from the cube, or not processed at all if they did not change.

If you never delete items from dimensions it may make sense to process UPDATE instead of full. (back to point one about knowing how it works)

3. Schedule processing differently.

Parallel processing may be killing the server. Try to process different things separately.

4. Configure SSAS to use more memory (memory limits in settings).

5. Taking the above into account it might make sense to do UPDATE processing mon-fri and FULL on sat-sun when noone is at work.

There is heaps of ways to improve

M.Sc.IT, M.B.A, MCTS BI 2008, MCITP BI 2008, MCTS SQL Dev, CSM, CDVDM

maheshsahini SSC Journeyman Points: 94 More actions · Answer 2

Thanks a lot Valek,

that was really helpful.

there are lot many aggregations and so have decided to go for an MOLAP storage.

I would take your advice on this and process the dimensions and measure groups separately, and also doing a full process once or twice a week would be a wise decision. Rest of the days I could do an update on dimensions that change.

renjith_githu Grasshopper Points: 24 More actions · Answer 3

Hi Both

I am facing an issue from last 20 days .For the last 2 years my cube was processing on an average time of 35 min.But last 20 days the processing time of the cube not in consistent.some day it i staking 3 hrs to process and some day 4 hrs .last 10 days 2 times it processed in 35 min too.Don't know why suddenly it is behaving like this.Any clues?

Note: all the ssis packages are completed on the same time as before

Thanks a lot

Renjith