I work in the Support/Operations area in our company and am the Admin for Power BI and am researching setting up new gateways.
We have many Preminum sites and are adding more and I understand the difference between the Gateways and Premium sites. This is all about the Gateway application running on the VM servers in a cluster. The groups that are using Power BI are building many very large datasets ranging from 6M to 600M in size. Some of them are refreshed once a day while others are refreshed every half hour. In the morning especially and occassionally during the day some datasets are getting timeout errors and failing and during those times there are large amounts of reports updating. When the same reports run later with fewer datasets refreshing they never fail.
So again, I am building new VMs and am clustering and I want to give the VMs as much power as I can but not go beyond the gateway applications ability to handle. I am begining the understand that there doesn't seem to be any documentation or testing of the gateways maximum abilities which surprises me having been a client/server application developer in my past.
This all goes alone with the lack of monitoring tools for Admins to watch the gateways closely and see what datasets are refreshing and causing issues using too many resources or running for long periods and reporting when there are too many datasets refreshing.
So my question to the community is what is the Max CPUs and Memory you have given your servers that was effeciently being used?
