Warren Kumari <warren@xxxxxxxxxx> wrote:
> There are dire proclamations that IPv6 TCP anycast cannot work.... And
> yet there are a bunch of existing implementations where it is clearly
> working, and people have built their business models around it, showing
> that it is working fine.
It can't work if the anycast mechanism depends upon flow-label remaining the
same during the lifetime of the connection. It seems that the BGP anycast
that I'm familiar with (as used by many anycast name server systems) can
work, and does work.
Maybe local load-balancers are borked. There was a talk at netdev about
trying to make the Linux load balancer code more stateless in the face of
hash bucket rebalancing, and I think that work is probably can never work.
I've ignored half of this thread, particularly the half where the quoting was
borked. It seems to me that there is a v6ops BCP that should get written.
--
Michael Richardson <mcr+IETF@xxxxxxxxxxxx> . o O ( IPv6 IøT consulting )
Sandelman Software Works Inc, Ottawa and Worldwide
Attachment:
signature.asc
Description: PGP signature
