Scott Brim - le (m/j/a) 4/1/09 5:30 PM:
Excerpts from Rémi Després on Wed, Apr 01, 2009 04:23:04PM +0200:
One problem with this name is that a NAT66 device is _not_
stateless. It requires configured/static state to function -- at
least the two prefixes between which it is translating. When I
say that the mapping is stateless, I mean that the mapping
mechanism does not create or require dynamic (per-host,
per-connection, etc.) state.
This is the kind of confusing situation I wish to help avoiding: "it
is stateless... but not stateless the way you thought".
How would we expect people to understand this outside a small
initiated circle?
By avoiding 'stateless'. What Margaret said.
Margaret and Fred also say, in the abstract of their draft, that they
describe "a stateless, transport-agnostic IPv6-to-IPv6 Network Address
Translation", in brief a stateless NAT.
This makes IMHO a lot of sense: this is what it is.
NAT as traditionally conceived has per-host (and maybe per-session)
state. NAT66 has per-prefix state. I think it would be fine if you
found a good name that reflected that, but did not make any claims of
being state-free.
Consider also that RFC 2765 defines the "Stateless IP/ICMP Translation
Algorithm (SIIT)" with more static parameters for address translation
than the two prefixes of the stateless NAT66.
I hope that, in view of these facts, you could accept that the word
stateless, rather than being avoided, should be used to keep things
clear, consistent, and simple.
Regards,
RD
_______________________________________________
nat66 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nat66
