Data flow not working

William Le

Dear all,

 

I want to make a data flow action with 3 cubes.

1. Cube A: 5 dimensions: month, base article, customer, production article, production manufacture.

2. Cube B: 3 dimensions: production article, production manufacture, sub article.

3. Cube C: 6 dimensions: month, base article, customer, production article, production manufacture, sub article.

All dimensions are sparsity except month. I want to compute C= A*B. Below are step what i'm doing now

C = A, with open sparsity

C'=B, with open sparsity. It takes a lot of time at this step, maybe because cube B only has 3 dimensions but C' has 6 dimensions.

C = C*C'.

Board takes a lot of time to calculate this. I think because i use data flow with open sparsity and huge cubes. Does anyone have any solutions for this case.

 

Thanks and best regards

Willam Le

Brendan Broughton

Hi William Le , I'm not sure what's in the cubes and if this would work, but can you avoid the C'=B, and do C'=A (w/ open sparsity) and then C = B*C' ? 

 

The other thing I'd generally look at when my dataflows are taking an inordinately long time is the selections that are present when the action is called.  Some of your cubes are by month, have you limited the months selected when the dataflow is called?  If not and your DB goes into the future then you'll also be calculating on these future periods which most likely don't have data.  The same goes for historical periods where the data may not require re-calculating.

Sometimes I will also use one of the source cubes in a dynamic select on one of the dimensions to try and spedd the calculation further, but this will depend on the sparse elements.

Björn Reuber

Hi,

 

calculation like this normaly take a while, cause the 3 dim and the 5 dim cube have to be enlarged (internally) to 6 dimension.

Cause of this reason our Dev Team devoleped the Join feature: Dataflows : c=a*b versus c=join(a*b) (currently in beta status, but available in actual versions). So you can use this feature to speed up your procedure. And join can also direclty create a sparsity so it should solve your problems

 

regards

Björn

William Le

Hi,

 

I tried to use Join feature, but nothing happen. I use Board version 10.1.3.

Thanks and best regards

 

William Le

Björn Reuber

Hi William Le

 

have you checked the #high performance mode?

Regards

Björn