cancel
Showing results for 
Search instead for 
Did you mean: 

SAP Data Intelligence - Modelling process take very long time.

0 Kudos

Hi,

I run fbprophet modelling on modeler (python3 operator) in SAP Data Intelligence. Normally on my local and on SAP DI Jupter notebook it only takes about 1 minutes. But when I deployed it on modeler it takes very long time (already run about 1 days and it is not finish yet). Maximum node for the SAP Data Intelligence was set to 8 node. I set the resource on the graph and the group. CPU: 4 core, Memory: 16Gb but the process only takes 0.6 core and 3Gb Memory.

How I can increase the resource usage?

Vitaliy-R
Developer Advocate
Developer Advocate
0 Kudos

Hi Agus. I assume the graph in Modeler is in "running" status, but do you have an indication that your operator received a trigger to start processing of the data? That's the first thing coming to my mind. Regards. -Vitaliy

0 Kudos

Hi Witalij,

Yes, the operator received a trigger. I also put some logs on my python script. All the process is working till stuck on model fitting. I believe it's running but take very long time. I think this is because it is only takes 0.6 core and 3Gb Memory.

Thank you.

jimgiffin
Product and Topic Expert
Product and Topic Expert
0 Kudos

Did you try increasing the memory and cpu under the graph resources? I'm not sure if fbprophet is single threaded but you could increase it to 1.0 CPU?

0 Kudos

Hi James,

Yes, I have set the resource on the graph and the group as well. I grouped the python operator. So there are resource request and limit.

Thank you.

Regards,

praveenctp
Explorer
0 Kudos

Any luck on mapping resource request and limit on graph pipeline? I do facing the same issue on on premise version of SAP DI. Any best practice reference?

Accepted Solutions (0)

Answers (0)