Troubleshooting and profiling datamapper and templates to increase performance

OL Connect Professional
1

First question

When profiling a Datamapper the performance seems to be measured once, aka the Datamapper steps are run and measured once. The template profiling instead performs several iterations. Why is that?

Dataming profiling

For instance this is the profiling of a datamapper configuration. The major offender is “Popola tabella di dettaglio” which indeed loops a detail table and for each entry queries an API hosted on the same machine of the OLConnect server, so ping is practically 0.

image
image861×483 88.2 KB

DATA
The API latency is around 2 milliseconds to process the request and give the reponse needed to fill some additional fields in the detail table.

The step took 333 milliseconds to complete, the data source had 10 entries, so 10 roundtrips were made for a total 20 ms process time on the backend side.

Can it be that the remaining 313 ms were to simply perform 10 addRow()?

Executing the datamining through the OLConnect API has a 900% time increase

Regardless of this profiling which overall adds up to roughly 500 milliseconds (still an excellent performance!), when actually performing the datamining step through the OLConnect API it takes up to 3 seconds to complete, how so?

The API being used is Process_Data_Mapping with a datamining configuration and a file already uploaded to the filestore.

I am measuring only the time it takes from the Process Datamapping POST to the the Get Progress of Operation 100%. Why the performance measured from the Designer Datamapper are so much different from the actual execution time through OLConnect API?

Template

The template profiling is unclear for me, it performs 1000 iterations, I guess to make an average.

Is the Elapsed time the cumulative time of all iterations or is it already the mean of the 1000?

Furthermore, why the first script “dati dinamici esterni” is executed 4 times more?

image
image829×596 65 KB

This template when executed through the OLConnect API takes about 6 seconds to complete.

2

About the “Template” part:

For template profiling there are tooltips that show more information.

If you hover over the 4000 value you should see why the script runs multiple times. In general, scripts run against each section in the active context and against each applied master page. If a script selector has no results for a particular resource the script will not run. If a script is meant to run against one specific resource it may be more efficient to place that script in a folder that is scoped to that resource.

The elapsed time is cumulative. It performs 1000 iterations by default (configurable in the preferences) because the measurements are too unreliable if the sample size is small.

3

Thanks! That will surely help indeed, I wasn’t aware of the feature.

I scoped two scripts to specific section and master page and had a delta of 7451 ms.