On Tue, Sep 20, 2022 at 4:12 PM Scott O. Bradner <sob@xxxxxxxxx> wrote:
so it does not require any changes to the software in end nodes or in routers and firewalls?
if it requires any changes then its a new protocol
Is Quic a new protocol?
I can solve the IPv4 address exhaustion problem with a small change to QUIC, a new DNS record and some changes to NAT boxen.
Observation, the only reason NAT doesn't serve is that a device behind a NAT cannot accept inbound calls.
The basic idea is:
1) Run everything over UDP
2) First n bytes of the UDP payload is an address extension specifying the destination.
3) The APLUS DNS record specifies an IPv4 address plus the prefix required to reach it.
4) Make the NAT box responsible for routing to the correct inbound endpoint.
5) Reserved IP address in outbound messages is used for signalling and to open up the external port.
There is some impact on applications but far less on the system than IPv6. The only forklift upgrade is to the carrier grade NAT. Everything else can be done inside applications.
Perhaps if instead of treating every alternative proposal as a threat, we could develop an IPv4 extension strategy that is also an IPv6 transition technology. The fundamental brokenness in the IPv6 transition strategy is that an IPv4 device cannot reach an IPv6 service.
It's been what, 30 years now? At what point does anyone else get a chance to try?
