Re: Another proposal

[sytek![e-mail] (Keith Mc Cloghrie)]

Jeff,

> [Here my proposal starts getting a little fuzzy.  Suggestions welcome.]
> 
> Presumably, we don't want to send it more often than once per RTT.
> That means that we should not send a second MTU Request Option until
> (1) we are asked to retransmit that datagram by the transport level
> (which presumably is tracking the RTT) or (2) we receive a packet
> from the destination host which is in reply to our original packet.
> I.e., we should use our existing RTT information in order to meter
> out the MTU Request Options.
 
I'm not sure the idea of using an RTT estimate is a good thing.
The time when you most want to get the information back, is when 
the route just changed, but if the route just changed, there's a 
good chance that the RTT also just changed.  So any RTT estimate
is not a good thing to use.  (I also dislike IP having to ask the
transport protocol what the RTT is, especially if it's UDP !!).

For discovering MTU increases, I suggest you don't need to retransmit.
Assuming that the caching of MTU is done in the IP layer, then MTU
increases only occur on the first use of a path after reboot and
on route-changes.  Route-changes are infrequent (hopefully reboots
are also !) and keeping the datagrams at the same smaller-than-path-MTU 
size for a little longer (i.e. until the next periodic request is sent)
doesn't do any harm; it just delays the good.

For discovering MTU decreases, see my suggested "report-N-fragmentations"
in my last message, in which I addressed the issue of loss.

> Next, how often should we send the MTU Request Option if the last-hop
> gateway doesn't understand it and simply isn't replying?  Easy
> approach: once per RTT.  It's mostly our own ox being gored by the
> cost of the excess option-processing.  Harder approach:  the last-hop
> gateway should normally change the option to indicate to the
> destination host that the ICMP has been sent.  If the option is not so
> marked, the destination host could send the ICMP itself.  (I think this
> is too finicky.  We can afford a few extra option transmissions.)

Are you here suggesting that the last-hop router and the destination 
host both send replies (or ICMP messages) ?

Your mention of the last-hop router not understanding raises
a different disturbing thought.  If both end-hosts and some (but not
all) of the intermediate routers understand the option, how do 
the hosts know that they can use the "discovered MTU", i.e. what
if the routers, which don't understand the option, are the ones at 
either end of the link with the smallest MTU ??  
 
Keith.