> On Nov 3, 2015, at 3:41 AM, Ebersman, Paul <[email protected]> > wrote: > > > On 03Nov, 2015, at 5:31 PM, Alain Durand <[email protected]> wrote: > >> In the particular case of the communication between the CPE and the ISP >> DNS recursive resolver, the two parties are within the same administrative >> authority. Thus, the need to make a BCP is much lower. This can be seen >> as simply an implementation issue. In other words, there are other >> solutions that could be used, for example a translation of the DNS packets >> from IPv4 to IPv6. Such a translation may or may not be optimal, but it >> would work and, more importantly, would not break the DNS resolution and >> would have no impact on the stability of the DNS system as a whole. > > Putting in a second DNS server that does nothing but forward everything just > to > translate v4 to v6 does indeed have an impact on stability if you try to do it > at large scale. It impacts infrastructure costs, performance and potentlally > confuses > geo-ip/cdn. It also adds complexity in debugging.
I was talking about doing the translation in the CPE. But, the larger point is, this was just an example of something else that could be done, although, I agree, sub-optimally. What draft-jjmb-sunset4-dns-forwarding-ipv4aas-00.txt proposes is reasonable, no objections. However, as far as I know, this is already being documented by the various IPv4aas solutions. Alain.
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ DNSOP mailing list [email protected] https://www.ietf.org/mailman/listinfo/dnsop
