I like the new organization of the document, and I generally support
its adoption. It could use a bit of proofreading and word-smithing,
but I have only one substantive objection.
Section 5.5 continues to say "the B4 element MUST implement a DNS
proxy". This is directly at odds with the new more relaxed (and more
complete) language in section 4.2: "it SHOULD operate a DNS proxy to
accept DNS IPv4 requests from home hosts and send them using IPv6 to
the service provider DNS servers".
I fully understand the reasons for operating a DNS proxy in the HGW in
general, and in the B4 HGW in particular. However, there are multiple
alternative DNS scenarios, none of which affect the operation of the
DS-Lite protocol:
1) The HGW could operate a proxy, using a statically-configured IPv4
resolver address (a third-party resolver, or even a resolver in the
provider's network, learned through out of band means).
2) The HGW could not operate a proxy, but pass along a statically-
configured IPv4 resolver address (as above) to its DHCPv4 clients in
the home network.
3) The HGW could operate its own recursive resolver, either over IPv4
transport or IPv6 transport.
4) Regardless of what the HGW does, a client in the home network could
be statically configured to use a third-party resolver, or to operate
its own recursive resolver.
In all these cases, the DNS queries and responses would be tunneled
through the AFTR like any other ds-lite traffic. As such, it places a
marginally higher load on the AFTR (something operators might want to
avoid), and in most cases requires static configuration (something
else operators might want to avoid), but it is perfectly acceptable
from a protocol point of view.
This is a long-winded way of saying that I strongly feel that the MUST
in section 5.5 needs to be changed to a SHOULD.
Proofreading edits to follow shortly...
paul
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires
