UpdateDetector much slower in FME 2017

Question

For a customer I had to exchange the FME 2015 UpdateDetector in a workspace with the FME 2017 version of the same transformer.

I have copied the same settings from the FME 2015 transformer and use the same dataset.

The performance for the workspace went from 5 minutes to 50 minutes, and a lot more (75%) updates were detected.

Has anybody the same experience with this custom transformer.

Just wondering what could cause changes to be detected with the new transformer that were not detected with the older version.

Any suggestions are appreciated.

icon

Best answer by erik_jan 21 November 2017, 22:29

erik_jan · Accepted Answer

The issue seems to be with the comparison of an attribute that has a different data_type in the sources:

Source 1 (Oracle) is Number(20,8), Source 2 (SDE) is double.

Even though the values are exactly the same, the records are detected as updates.

At the end it is the updates that are slowing down the workflow.

erik_jan · Answer

The issue seems to be with the comparison of an attribute that has a different data_type in the sources:

Source 1 (Oracle) is Number(20,8), Source 2 (SDE) is double.

Even though the values are exactly the same, the records are detected as updates.

At the end it is the updates that are slowing down the workflow.

Fixing the data_type issue solved the workspace performance issue.

Issued a call to Safe Support about this.

Sign up