Dataflow process error

Hello for a few days our dataflow is crashing with this error message

Dataflow process error -->
Error: Compute engine failed: Insufficient memory. Please increase your compute engine/workload memory and try again.
(8645). Details: A timeout occurred while waiting
for memory resources to execute the query in resource pool 'default' (2). Rerun the query.

Have you ever experienced this error message , please?

We use a non-clustered on premise gateway to access the source data
We are looking for memory stress due to treatments
(40 entities in the dataflow) and install a cluster
But maybe it's due to the volume

Thank you for any help and recommendation

Best regards 

Christophe