[PATCH] kern/681110 re-roll of RFC3522 (Eifel detection) patchset

[Mark Allman]

> BTW. Is there any legal alternative, e.g. make this as an option like
> WANT_IDEA?

I do not know what WANT_IDEA is.

As far as alternatives, there is F-RTO (RFC 4138) and DSACK (RFCs 2883 &
3708).  We have also worked on a response that is different (and
unencumbered).  It is just an internet-draft (and, in fact, needs rev-ed
because we have a better idea in mind):

  Josh Blanton, Ethan Blanton, Mark Allman.  Using Spurious
  Retransmissions to Adapt the Retransmission Timeout. December
  2006. Internet-Draft draft-allman-rto-backoff-04.txt (work in
  progress).  

The problem is that these are not technically as good as Eifel, IMO.
But, Ericsson has taken the obvious way to do things off the table, I am
afraid.  (Lots of the Eifel work would seem to me to not stand up to a
challenge based on previous/simultaneous work and the obviousness of
it.  But, that is clearly pure conjecture and I am not putting up the
funds to test the theory.)

The good news is that in most environments if you use the standard RTO
timer (RFC 2988) then the number of spurious timeouts experienced is
vanishingly small.  See:

  Mark Allman, Vern Paxson.  On Estimating End-to-End Network Path
  Properties.  Proceedings of the ACM SIGCOMM Technical Symposium,
  Cambridge, MA, September 1999.
  http://www.icir.org/mallman/papers/estimation.ps

allman



-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 185 bytes
Desc: not available
Url : http://lists.freebsd.org/pipermail/freebsd-net/attachments/20070412/2e671ce2/attachment.pgp