Speed of TX REST vs CData REST

Are you testing on the same VM or a VM with the same specs?

To increase performance please try enabling incremental load

Running on exactly the same VM.

Do you have any instructions/documentation on how to set up incremental load on TX version 20.10? All I get when trying to set this up is “Incremental load is not supported on this table”.

Hi ​@sten.lomme 

How did you test the speed? Have you tried to do this in the newest v20 release 20.10.74?

Was the calls exactly the same?

Hi. We set up connections to the same API endpoint with similar setup using the new connector. Running on the same dataset. Tested multiple times to prevent other executions from interfering with process power. It is a very small endpoint, but increased run time was aprox 4x. So I wanted to know other peoples experience on this. For a bigger endpoint this increase would be dramatic.

In your Classic support there is a ticket with the XSLT and RSD.

Hi,

are the data types detected by TX identical between the CData and TX enhanced connector?

Have you enabled logging? That may affect performance (but not that much).

No, they are detected as different data types.

In the faster CData source, both fields are nvarchar(2000).
In the slower TX Enhanced, both fields are INT.

In this case would expect the opposite result.

I have tried both with logging enabled and disabled. Seems like the same result.

Hi,

CData would typically just apply (n)varchar(2000), but in some cases the TX Enhanced connectors choose wider types which can sometimes explain extra slowness / resource usage. In this case I would expect int to be more efficient.