Re: [mdnsext] mDNSext features/requirements rollup

[Michael Richardson <[e-mail]>]

>>>>> "RJ" == RJ Atkinson <[e-mail]> writes:
    RJ> Of course, we ought not REQUIRE that it be implemented in
    RJ> routers -- both in the sense of not requiring router
    RJ> implementers to include it and also in the sense of not
    RJ> requiring network operators to put the function in routers.

+1

Perhaps we should write that this function must be easily co-located
inside a router, but that it should not depend upon any (default)
routing infrastructure to actually bring packets there.

I can imagine one implementation where we would define some anycast address
to which DNS UPDATEs and queries should be sent to, and we would assume
that the routing infrastructure would respond, with the result being
blackholed by ISPs.  Such a system would *require* the proxy function to
be co-located in routers, which I think is what you are getting at: we
don't want to require it.

A system of proxying link-local multicast would not require co-location
the way an anycast would.

-- 
]               Never tell me the odds!                 | ipv6 mesh networks [ 
]   Michael Richardson, Sandelman Software Works        | network architect  [ 
]     [e-mail]  http://www.sandelman.ca/        |   ruby on rails    [