Do not input private or sensitive data. View Qlik Privacy & Cookie Policy.
Skip to main content

Announcements
Join us in Toronto Sept 9th for Qlik's AI Reality Tour! Register Now
cancel
Showing results for 
Search instead for 
Did you mean: 
ganeshK
Contributor
Contributor

tDB2Output Component Parallel Execution issue

Hi Team,

I have job with tDB2Output component for Data insertion. In Advance settings i have tried enabling Enable Parallel Execution = 2 , getting poor performance compare to normal execution.

When i enable this option

no.of records insertion is only 9-10 records per second and CPU performance is good 0.2-0.3%. In normal execution without parallel execution option no.of records is about 70-75 records per second but CPU utilization is high around 2.4 - 3 %

Enabling Parallel Execution : 9-10 records per second , CPU Usage - 0.2 - 0.3%

Without Enabling Parallel Execution :

70-75 records per second ,

CPU Usage -

2.4 - 3%

Can someone guide me how do i overcome this issue. What is the main benefit of Enable Parallel execution in this component.

My Job Design flow:

tFileInput -> tDelimitedInput -> tMap -> tDBOutput (IBM DB2) -> tDBCommit (IBM DB2)

0695b00000G3I6XAAV.png

Regards,

Ganesh.R

Labels (2)
3 Replies
Anonymous
Not applicable

Hi

What is the action on data option? The 'Enable parallel execution' option is usually used for data insertion. Use an existing DB connection on tDBOutput? Only tDBCommit component, no tDBConnection in the Job Design flow?

 

Regards

Shong

ganeshK
Contributor
Contributor
Author

Hi, Data option is "Insert". Please find the job design flow.

 

0695b00000G3Y9CAAV.png0695b00000G3Y8xAAF.png 

sensiva
Contributor
Contributor

Hello,

 

I am facing the same issue, with parallel execution option, the job is very very slower than the single thread execution. I initially started with 2 threads which was less performant than single thread execution and then i switched to 4 threads and the result was even bad. Just wondering, if there could be a bottleneck somewhere..