resistance [to ipv6] is futile - http://www.spreadshirt.com/-C3376A12786302
I certainly don't think IPv6 will reach 100% deployment, people
will continue [to get paid?] to operate 6to4 and 4to6 gateways, even if just
enterprise edge from a nat44(+) gateways.
I'm still waiting for a few orgs to make the IPv6 jump like
Wayport/attwifi as an example. They could do nat66 like they do nat44
and easily make the sites look the same through their templates.
If you assume most people are right and Netflix is 33% of the US
internet at peak, that's 33% that's fully ipv6 capable on the server-side.
facebook, google and others count up as well, so much of content is reachable.
If apple/icloud make the jump to publishing AAAA records as part
of their CDN efforts to move away from akamai, I suspect much more of the
LTE traffic would make that jump to v6. (Waiting to see ATT upgrade their LTE
to support v6 to match VZ).
It also appears that OSX 10.10 fixes some of the IPv6 issues that exist
in 10.9, so with that update in the coming months I'm expecting even more IPv6
traffic to replace the IPv4 bits.
- Jared
There would be a root, or multiple roots, which would respond to
requests to locate who should be asked about a domain, for example if
you want to know the ip address for world.std.com the conversation
goes roughly:
(To Root Server): Where is the COM server?
(From Root Server): SOMEHOST
(TO SOMEHOST): Where is the STD.COM server?
(From SOMEHOST): 192.137.74.112
(TO 192.74.137.112): WHAT IS WORLD.STD.COM's IP ADDRESS (A RECORD)?
(FROM 192.74.137.112): 192.74.137.5
Not quite right. It actually goes like this on the wire:
(To Root Server): WHAT IS WORLD.STD.COM <http://world.std.com/>'s
IP ADDRESS (A RECORD)?
(From Root Server): I don't know, but SOMEHOST is the one to ask
about COM
(TO SOMEHOST): WHAT IS WORLD.STD.COM <http://world.std.com/>'s IP
ADDRESS (A RECORD)?
(From SOMEHOST): I don't know, but 192.74.137.112 is the one to ask
about STD.COM
(TO 192.74.137.112): WHAT IS WORLD.STD.COM <http://world.std.com/>'s
IP ADDRESS (A RECORD)?
(FROM 192.74.137.112): 192.74.137.5
Or the DNSSEC option:
(To Root Server): WHAT IS WORLD.STD.COM <http://world.std.com/>'s
IP ADDRESS (A RECORD)?
(From Root Server): I don't know, but SOMEHOST is the one to ask
about COM, and you can trust SOMEONE if it signs with COM-Key. Signed with
ROOT-Key.
(TO SOMEHOST): WHAT IS WORLD.STD.COM <http://world.std.com/>'s IP
ADDRESS (A RECORD)?
(From SOMEHOST): I don't know, but 192.74.137.112 is the one to ask
about STD.COM, and and you can't tell whether you are really talking to
192.74.137.112 since it's not signed. Signed with COM-Key.
(TO 192.74.137.112): WHAT IS WORLD.STD.COM <http://world.std.com/>'s
IP ADDRESS (A RECORD)?
(FROM 192.74.137.112): 192.74.137.5.
Rubens
Barry Shein wrote:
The idea is very simple, each site would be responsible for their own
domain and to respond to simple remote requests for name to ip address
mappings or back again.
Wrong. DNS is not that simple.
Domains and sites have, in general, independent topology
that sites can not be responsible for domains.
Perhaps, your misunderstanding is commonly shared by those
who believe in NDN, though they might think there are
negligible number of exceptions.
Then, data, mostly, could be routed based on name hierarchy,
which scales well.
The reality, however, is that exceptions are everywhere
and we need something like DNS to translate names into
something scalably routable, that is, hierarchical
addresses.
Masataka Ohta
Understand these were speaking notes and it was safe to assume the
audience basically understood DNS so it wasn't my intention to give an
exhaustive introduction to how DNS works.
There also seems to be some splitting of hairs over the meaning of
"site" in your response. That is, some sort of physical boundary vs an
authoritative boundary.
At any rate my proposal doesn't eliminate hierarchical addresses, it
just says (in brief) that "bits is bits" and IP numeric addresses per
se were mostly a product of modeling fast CPU registers which may not
be the only model. One could use the FQDNs themselves as hierarchical
addresses at least as an external representation.
It was intended to be a provocative proposal.
Barry Shein wrote:
Understand these were speaking notes and it was safe to assume the
audience basically understood DNS so it wasn't my intention to give an
exhaustive introduction to how DNS works.
Surprisingly many people who basically understand DNS have the
same misunderstanding as you, which is why some people believe
in NDN.
There also seems to be some splitting of hairs over the meaning of
"site" in your response. That is, some sort of physical boundary vs an
authoritative boundary.
Then, "site" based FQDN can not be used for scalable routing.
At any rate my proposal doesn't eliminate hierarchical addresses,
See above.
One could use the FQDNs themselves as hierarchical
addresses at least as an external representation.
You are trying to define something not usable for scalable
routing a hierarchical address, which is as bad as your
attempt to distort the definition of "site".
Masataka Ohta
Well, it's a good thing we have you around to keep us honest.