Networking/spanning tree loops, arp table mismatch/corruption, the switches at the destination being misconfigured are all realistic problems that would result in downtime here. The normal way you do this is with live migration from hyper-v or vmotion from ESXI. If the initial migration is not successful, you just leave the server powered on while you address the issues. Once the VM has been migrated you can do whatever you want with the original server without worrying about downtime.