[Dnsmasq-discuss] High Availability: Part Deux

[Dave Taht]

I too would like a more high availability form of DNS and dhcp in general.

One thing that I do currently is use anycast in my (fairly complex,
highly routed) campus network, so that the local dns servers are
distributed via the babel routing protocol, and the closest one that
is up responds. (anycast is widely used on the internet backbone as
well)

this introduces problems in monitoring what, exactly, is "up", and is
not as effective as what I would like. Replicating the dhcp leases and
dns entries remains a problem (what usually happens is users trying a
non working AP, end up trying another).

I still regard this distributed solution as it stands, as being
superior to the alternative of hauling back all dns and dhcp to a set
of central servers, which then would require perfect connectivity at
all times, centralized management of everything, and result in worse
caching.