At 08:07 PM 10/17/98 +0000, Michael Shields wrote:
In article <4.0.1.19981016224901.00de6240@pariah.cncx.com>, I Am Not An Isp <patrick@ianai.net> wrote:
FACT: RFC1918 space does not break PMTU discovery. Deal with it.
If you use RFC 1918 space on the Internet, PMTU very well may not work for you. You can place fault on whatever standard you like but that doesn't change the *operational* issue.
Mark and I have discussed this in private. I believe this has devolved into an argument over semantics. Assume you use RFC1918 addressing in a private enterprise. PMTU-D will work without a problem within that enterprise (unless you filter your own routers). QED: RFC1918 does not break PMTU-D. However, if packets from RFC1918 ports leave the enterprise, then there is a very real possibility those packets will be filtered on other networks. These filters are in no way wrong, bad or otherwise a problem. It's just something you have to deal with "operationally". So, using RFC1918 addresses on router ports which have any possibility of send packets (even ICMP one-way communication) to other networks may find their packets filtered. This means that things like PMTU-D will break. Not using RFC1918 addressing reduces this possibility because (as Mark pointed out) these filters are recommended by the RFC. Does that make everyone happy? Can we get back to our regularly scheduled furniture, carpet and spam discussions? :)
Shields, CrossLink.
TTFN, patrick I Am Not An Isp www.ianai.net "Think of it as evolution in action." - Niven & Pournelle