Michael Krygier wrote:
Any comments about what Drplokta wrote on his blog about Windows Vista implementing RFC3484 and breaking Round Robin DNS?
It does *NOT* break Round Robin DNS, it breaks the assumption what people make when they setup their DNS in that manner. Slight difference ;)
Clients running Vista and where you have a Round-Robin DNS setup where the prefixes are very different will now hit the 'closest' prefix instead of a random one like most implementations of DNS resolvers, that is at first hit, when the connection to the first one fails they will fall back to the next closest one. If you have your hosts on the same subnet you won't notice the difference.
See the dnsops lists and other lists where 'Drplotka' spammed this too also.
This issue has been known already by the IETF IPv6 Maintainance WG (6man / http://www.ietf.org/html.charters/6man-charter.html ) see the list there for the discussion and it looks like there is a direction where people want to scrap that rule, which made sense at one point in time for IPv6 (though not for IPv4, which is a flaw in the RFC) but doesn't anymore.
Greets, Jeroen