[dns-operations] Anycast vs. unicast NS
David Ulevitch
david at opendns.com
Mon Mar 21 14:46:52 UTC 2011
On Mar 21, 2011, at 6:29 AM, Simon Munton <Simon.Munton at communitydns.net> wrote:
> On 20/03/2011 01:00, Stephane Bortzmeyer wrote:
>> Anycast is great for reliability and resiliency, but it does not
>> optimize the RTT.
>
> Our experience is the same - see my previous post on this thread on 18th
>
> One time we had queries from Russia ending up in Tokyo, going past our nodes in RIPE, London, Ashburn & LA - so not even the shortest route to Tokyo!
>
> We fixed it, and now have nodes in Moscow & StP
>
> is the only answer to this more & more & more nodes?
No, it's having consistent transit providers across nodes, judicious peering policies that don't trombone traffic and a mix of other network policies ( peering-only nodes, etc. ) that will prevent that from happening. And fwiw, the RTT is never worse than unicast would be, by definition, as the topologically far route is still their "best" route.
> _______________________________________________
> dns-operations mailing list
> dns-operations at lists.dns-oarc.net
> https://lists.dns-oarc.net/mailman/listinfo/dns-operations
>
More information about the dns-operations
mailing list