I like this idea in principle, but I am quite concerned by
two issues that seem to make it unattractive in practice:
1. No defined method for discovering the Subscriber IPv4 Address
(section 3.3) and no universal method of discovering the SP-BR
prefix (requires RADIUS or DIAMETER, section 5.2, but some ISPs
don't use either of them).
2. Does not support IPv6 server ports (section 7.1, no
externally initiated sessions). That's just as bad as traditional
NAT.
One detail in section 4.1:
> o this specification does not introduce new requirement to the IPv4
> UDP Length and Checksum.
Does that mean that the UDP checksum is optional?
Brian
_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires
