Implementing DHCP Fail Over/Load Balancing
As shown in previous recipes, installing and configuring a single on-premises DHCP server is straightforward. However, a single DHCP server represents a single point of failure, which is never a good thing. The solution is always to have a second DHCP server with as much independence as possible from the first server. That would include running the second DHCP service on a separate host (physical or virtual) on different subnets and using independent power and networking components.
In earlier versions of Windows, one approach to fault tolerance was to stand up two DHCP servers and define the necessary scopes on each DHCP host. You split the full set of IP addresses and allowed each server to have part of that set. The traditional 'wisdom' was to do an 80/20 split (have 80% of the scope supplied by your primary DHCP server and 20% on the backup server).
Independent DHCP servers are an error-prone approach and were never ideal since...
