The Atera Agent should be able to stay online (ie not drop off the console/portal and show as OFFLINE with red dot and raise events) if it loses DNS -- but it does. And thus so does the remote access with SpashTop (if you use that - we dont) and any other control of the agent.
I noticed tonight when I rebooted a local client server all their Atera Agents on the workstations also went offline for the hour it took the server to reboot, install Proliant firmware updates, windows updates and then come back online.- once the local server was back online the Agents "reconnected"
In a domain environment we cant use a public DNS server on workstations - they must use the domain controller for DNS (even issuing a public DNS as a secondary can cause issues and thwart troubleshooting)
Your server farms would be with Microsoft or Amazon or something that would be regional and the IP would not change too often... The Agents need to cache the last few IP addresses of their "backend servers" so they can use this if DNS is taken away from them.
All the popular remote access and RMM products do this so they dont drop offline if they lose DNS temporarily or can boot-strap themselves if they need to without DNS available... TeamViewer, Solarwinds, ScreenCOnnect etc stay online through a DNS outage and have a half dozen IPs to fall back on.
Another advantage of having the workstation Agents staying online while a server is down/unavailable is that may be the only way for techs who dont have multiple options of remote access or vpn of accessing the server/network in a outage - a jump box - just connect to a workstation on same network and then connect to the servers ILO/DRAC etc