IN-DB tools running slowly when run directly from the Alteryx Gallery

Question

We're troubleshooting an issue observed by one of our customers whereby their workflows are taking significantly longer to run from the Gallery than they do when run from Designer.

Our investigation has discovered the following:

* When workflows using IN-DB tools are run via the Gallery, they take up to 40 times longer to process compared to when run from Designer.
* This includes when the workflows are run from Designer on the same machine that Alteryx Server is installed on. i.e.:* Run from Designer on Gallery machine = normal IN-DB tool speed.
* Run from Gallery web UI = slow IN-DB tool speed.

* Customer uses 2023.1, however we've tested this on our own 2023.1 instance and we do not receive the same behaviour, so it is specific to their environment or config.

We're starting to investigate Run As permissions (could be defaulting to a sub-optimal driver when run from Gallery?) but any thoughts, advice, or experience on the topic would be incredibly useful here.

I believe there may be an Alteryx Support case for this but so far there hasn't been any obvious cause identified.

JonathanAllenby · Answer

Thanks for the input guys!

Using the evidence we gathered from our own environment we were able to get Alteryx Support to take a deeper look, with the outcome being that this is actually a new defect:

* "We are very sorry to see you are running into this issue. The issue you have reported has been reproduced and logged as a new Defect: GCSE-1493. This Defect will be reviewed by our development team. We will notify you of any updates we receive.
Workaround: Do not use DCM with the In-DB tool"

apathetichell · Answer

Two potential issues:

1) do you have a firewall that is specific to Server operations - this could require a level of tracing/protection/VPN routing which would make running a workflow through server take longer.

2) Do we know which machine the workflow runs on when run on Server ie which worker node? Do you know if there are substantial memory differences on the worker node (ie it's a 8gb machine vs a 32gb local machine and a 64gb server machine)... That could also slow down performance.

Which DB are you using btw?

fmvizcaino · Answer

Hi @JonathanAllenby ,

The only thing I can think of is related to the driver. The user using a user-DSN and the run as account using a system DNS.

I don`t really understand proxies, that said, another idea could be the run as account is configured to use a proxy and the user isn`t (this could be totally wrong)

If you have the run as account password, I would try opening the server designer with the run as user and run the workflow. (shift + right click)

Best,

Fernando Vizcaino