Comparing schedule refreshes from 2 gateways directly.

RE: Comparing schedule refreshes from 2 gateways directly.

Hello,We can propose a first version that we already use since months (first use 19/07/2023), after removing the sources and credentials.Here is our context:We have a fairly complex data refresh schedule that varies every day with a large number of data refreshes. We noticed that the refreshes took longer at certain times, and varied slightly from one day to the next (between 0:00 and 3:00 and then around 6:00), without any warning about the occupancy of the source systems or the on-premise gateway services.To analyze this phenomenon, we built 2 identical reports:- dataset type, then dataflow type (both are concerned)- to a single SAP Hana source (currently being upgraded to add a Sharepoint source, so without on-premise gateway flow)- The 2 Datasets and Dataflow are identical, same HANA view, same transformation in Power Query. The only difference is in the Gateway used for each refresh. In one case, we use the Gateway Prod, in the other the Gateway tech QUAL. The differences in duration between the 2 refreshes therefore depend on the Gateway, its network or its ability to communicate with the PowerBI.com bus. On the AWS PROD Gateway, there is a peak between 0 and 3 hours, which corresponds to the period during which this Gateway is "overloaded". During this period, refresh times can be multiplied by 10 or 15 on a few occasions between 0:00 and 3:00 and then around 6:00, every day, including weekends. No parameter monitored on the Gateway server confirms this overload, and in fact it has been resized twice, doubling the size, with no improvement and probably generating unnecessary costs. We need to identify the source of this limitation in order to find a solution.