Skip to main content
Archived

Idea - Discover performance bottle necks in workbench

Related products:FME Form
  • October 5, 2018
  • 4 replies
  • 88 views

Would love an easy way to show processing time PER transformer to quickly identify transformers that a performance bottle necks.

The number would be the processing time of features being processed by that transformer.

Either by a report, log or visually on the workbench when running in the feature caching mode.

This would be of great help to prioritize work on performance/workbench timings to schedule data processes on time.

This post is closed to further activity.
It may be an old question, an answered question, an implemented idea, or a notification-only post.
Please check post dates before relying on any information in a question or answer.
For follow-up or related questions, please post a new question or idea.
If there is a genuine update to be made, please contact us and request that the post is reopened.

4 replies

mark2atsafe
Safer
Forum|alt.badge.img+59
  • Safer
  • October 5, 2018

I do have an enhancement request already filed with the developers to do something like this. In my case it was to highlight each transformer as it is processed, but this is very similar. The reference number is FMEENGINE-47325

However... it should be noted that FME doesn't process data one transformer at a time. It processes data one feature at a time. Therefore it's hard to calculate exactly how much time is spent on one particular transformer, or which transformer is currently being processed. That's just to emphasize that this isn't a simple task to achieve, and that if we did implement this, it might cause a slowdown in processing. I would still like to see something like this, but I wouldn't expect it any time soon!


fmelizard
Safer
Forum|alt.badge.img+22
  • Safer
  • October 5, 2018

There was an FME World Tour talk showing how to open the developer back door on this: https://www.safe.com/presentation/fme-performance-profiling/ -- Results can be hard to understand but this is the foundation upon which we'd built a nice UI one day.


  • Author
  • October 8, 2018
I would view it like the feature cache, a tool you would use during "debugging", almost like put your workbench in debug mode. I think it would be acceptable that the time per transformer may truly not be accurate, you couldn't take into account what else was going on the machine at the time of processing. But the times of each transformer should be roughly relative enough that one could sort the worst to the best to then allow us to focus efforts on improving performance. I wouldn't use this at the scheduled times, more when I'm building and testing.

 

 

If I noticed a poor performing workbench one day then I could take it off into debug mode to figure out what's going on.

 

 


david_r
Celebrity
  • May 4, 2020

See also the suggestions FMEENGINE-60629 and FMEENGINE-60631 that I submitted at the 2019 Partner Summit, there's a bit of overlap with some ideas here.