simple hosts/routers and MTU [Re: Path MTU for draft-ietf-v6ops-mech-v2-00.txt]

[Pekka Savola <pekkas@netcore.fi>]

On Tue, 6 May 2003 itojun@iijlab.net wrote:
> 	and, i don't think you have convinced me on the technical benefits
> 	of 1380 (the particular value).

I'm trying to look at this from a different angle: what benefits would
1280 really have (except for simplicity etc.etc. vague things)?

To me, it seems that the features possibly gained by pushing 1280 
could be:

 - can hosts be implemented so that they do not implement fragmentation 
and reassembly? (no difference, it seems)
 - can hosts be implemented so that they don't need to participate 
actively in PMTUD? (no difference, it seems)
 - can hosts be implemented so that they don't need to care for received 
"packet too big" ICMP messages ("passive PMTUD")? (passive PMTUD would 
seem to be required if the host uses 1380 for sending)
 - can hosts be implemented so that they can have a lower MRU than
1500 bytes? (no difference, it seems)
 - ...

As for routers,
 - can routers be implemented so that they do not need to implement 
IPv4 fragmentation and reassembly *for tunnel packets* (note: IPv6 frag + 
reass may be needed anyway, from "host part")?
 - can routers be implemented so that they can have a lower MRU than 1500 
bytes? (can't implement that way, I think.)
 - ...

Please try to come up with some concrete thoughts on this?
(I'm not sure whether 1380 is the best choice anyway, but I'd like to get 
some harder data to back up some rather abstract reasons.)

-- 
Pekka Savola                 "You each name yourselves king, yet the
Netcore Oy                    kingdom bleeds."
Systems. Networks. Security. -- George R.R. Martin: A Clash of Kings