> Tony Finch <mailto:d...@dotat.at> > Wednesday, March 25, 2015 7:31 AM > k...@wide.ad.jp <k...@wide.ad.jp> wrote: >> I don't know if such ISPs still exist, but when we started Anycast, it >> was reported that some ISPs did per-packet load balancing. > > If they do that they will utterly wreck the performance of unicast TCP, > which generally does not cope well with out-of-order packets.
and yet, in a UDP-first UDP-mostly or even UDP-only operating environment, that assumption has been sound. if we're about to make it unsound then we have to say so. > >> The root zone grows over time in almost monotonous manner and longer TCP >> sessions (especially over narrow bandwidth circuits) could break due to >> routing change. > > That's also true for unicast TCP since BGP's typical convergence times are > not much longer than TCP connection timeouts. there is no convergence any more. the diameter and surface area of the bgp cloud has for some years now been such that some packets always go awry (either routed suboptimally, or sent to the dead packet office). short tcp sessions which don't go idle or for which responder-close is a well-understood and widely-expected condition, are not affected by this. axfr of the root zone, even at current sizes before the monotonous growth kato has described, is not a short enough to fit this envelope. i think the document has to mention this. -- Paul Vixie
_______________________________________________ DNSOP mailing list DNSOP@ietf.org https://www.ietf.org/mailman/listinfo/dnsop
