Re: comments on draft-park-pppext-mip6-co

Bernard Aboba writes:
> > Why is a single round-trip time important?  Are there really links
> > where this round-trip time is significant compared to any network
> > layer traffic that might flow on the link?
> 
> The absolute value of the round-trip time may be significant on links such
> as GPRS or UMTS where RTT can be 800-5000 ms.  Network layer traffic will
> experience the same level of delay, however.

As Vern and others have pointed out repeatedly (and, yes, this is my
paraphrasing, so don't blame him for it; blame me instead): if a
single RTT is a heart-breaker, then you're just not going to be able
to use the resulting link for any reasonable networking-related
purposes.  Shortening the IPV6CP negotiation time doesn't help a whit
unless you really think you're going to bring the link up, send out
one packet, and then slam it back down again.

So, the question still stands: in what scenario is the connection
set-up time an _appreciable_ fraction of the time spent conversing
with network peers, such that eliminating a single RTT will actually
result in some noticeable and worthwhile benefit versus the disruption
that adding (and deploying) yet another option will incur?

> > Is this perhaps papering over implementation flaws, such as an
> > inability to buffer datagrams when the link is down or (maybe)
> > triggering retransmit at opportune times?
> 
> The high RTT times in GPRS/UMTS are in part due to large buffers in the
> GGSN;

Perhaps I don't understand some crucial part of the architecture here,
but shouldn't those buffers be essentially empty (and delay minimized)
at the point where we're still just bringing the link up?

>  today's TCP implementations may cause those buffers to fill,
> resulting in very high RTT values, and somewhat unstable TCP performance.

Sounds like IPV6CP is just about the least of the worries.

-- 
James Carlson, IP Systems Group                <[email protected]>
Sun Microsystems / 1 Network Drive         71.234W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.497N   Fax +1 781 442 1677

_______________________________________________
Pppext mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/pppext