Re: [AFMUG] DHCP Server Redundancy Survey

[Jesse DuPont]

Sterling, When we did DHCP, we did have a pair of centralized servers, both running ISC DHCP on Linux. The configs on the two servers have to be nearly identical (except for the declarations about the server ID and the server role) so I ran an NFS share between them. Each had a very small, local dhcpd.conf file with the unique portions and then an include statement for a common file that both servers used for all their prefixes. The routers would DHCP-RELAY to both servers and the servers had an algorithm to spread load. In this way, the same pool was defined on both servers and each knew whether the other served it or not. I did not use it for v6-PD, but did for v4 and it worked well. Failover worked very well, too. Yes, having the DHCP servers get lease approval from RADIUS would allow better northbound integration, assuming the billing system knew which MAC address belonged to which customer. Jesse DuPont Network Architect email: jesse.dup...@celeritycorp.net Celerity Networks LLC Celerity Broadband LLC Like us! facebook.com/celeritynetworksllc Like us! facebook.com/celeritybroadband On 10/31/17 1:59 PM, Sterling Jacobson wrote: I want to do a more flexible/standard setup for my DHCP handouts. I hand out public IPv4 and IPv6 dual stack to each customer router from two main routers on my network. Is it best to create two redundant DHCP servers instead and use DHCP Relay on-net to them? And how is everyone doing that? I'm guessing it's probably best to have those two redundant DHCP servers be RADIUS controlled so billing systems can easily integrate with them.