[why aren't my messages making it to lists? sending again...]
Jari, this should answer your message just now, I suggest we get some
people together and do some more testing in Dublin.
On 23 jul 2008, at 2:25, Dave Thaler wrote:
Both Windows vista and MacOS Leopard Dual stack connecting to
test1: it
works so it sends IPv4 packets... this seems reasonable, but i wonder
if it is always ok.
Both Windows vista and MacOS leopard Dual stack sending to test2: it
works and it they both send IPv6 packets, (this is what we want!)
Yes, this is excellent news.
Not really, because it doesn't buy us anything: it just means that on
dual stack systems, the mapped addresses in AAAA records don't get in
the way.
What we need is for IPv6-only systems to successfully send and receive
IPv6 packets with mapped addresses as the source or destination. And
we haven't seen any type of system do this so far.
I think you need to consider what scenario you're targeting NAT-PT
for.
I would argue that a dual-stack OS with IPv4 disabled is not a common
scenario. Hence while I agree that the OS's tested appear to be
problematic, it's not clear to me that that's a very useful data
point.
We were able to quickly test with IPv4 disabled, testing with IPv4
enabled but no IPv4 available would have taken considerably more time.
In theory no IPv4 means no IPv4, regardless of why, but there is the
possibility that this makes a difference, as I've seen some
applications on my Mac work over IPv6 with IPv4 enabled but not
available but fail with IPv4 disabled.
Remember that the case you're evaluating here is a dual-IP-version
network, not a v6-only network (I believe we already agreed the
question is moot in a v6only network).
What question?
NAT-PT/NAT64 inherently address hosts that only have IPv6
connectivity, probably because there is no IPv4 available even though
the hosts support it, but possibly because the host doesn't have a v4
stack.
If you have a dual-IP-version
network and a host OS that supports both IPv4 and IPv6, why would
anyone disable IPv4 on the host? That doesn't seem like a very
important case to me.
Right. But for a quick test it's much easier to turn off IPv4 for an
interface on a host rather than convince the IT department to turn off
IPv4 on the production network or go into the lab and set up an IPv6-
only network, so that's what we did.
I gather that in Dublin we'll once again have an IPv6-only network, so
perhaps we should retest there.
The real v6-only scenario you want to test is stacks/devices that have
_no_ IPv4 implementation at all, and you want to know whether those
treat v4mapped addresses the same way as any other native IPv6 address
when connected to a dual-IP-version network.
Do systems like that exist today?
Obviously we're trying to come up with something that will be useful
on systems that are in wide use within the next few years, which means
it must work on existing Linux, Windows and MacOS, if not on the
current versions then on versions that can be pushed out as
incrimental updates within a couple of years.
_______________________________________________
Int-area mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/int-area
