[CALUG] Gigabit Ethernet

[Jason Dixon]

On Oct 20, 2005, at 7:36 PM, Jim Bauer wrote:

> On Wednesday 19 October 2005 18:30, Ray Lischner wrote:
>
>> So why doesn't PMTU discovery work in this case? Why can't the jumbo
>> frame device discover the path MTU is only 1500 (or whatever) and  
>> send
>> frames accordingly when it needs to talk to that device?
>
> It can't discover the PMTU is only 1500 as it won't get
> the necessary ICMP error packet because the node the jumbo
> frame was sent to was incapable of receiving it.  You can't
> reply to what you haven't received.

That's not the proposed design of PMTUD, at least according to  
RFC1191 (1990).  The idea is that the router will find the minimum  
MTU for all hops, then issue the ICMP error to the source.  It's hard  
to tell what (if anything) has ever been successfully adopted by the  
market.  Everything I've read refers to very old RFCs, or proposed  
new solutions.  The IETF charter on PMTUD describes a new solution  
where progressively larger packets are transmitted until the  
"ceiling" is reached.

http://www.ietf.org/html.charters/pmtud-charter.html

--
Jason Dixon
DixonGroup Consulting
http://www.dixongroup.net