[LOOPS] LOOPS BOF

[Marie-Jose Montpetit <[e-mail]>]

My take on Monday’s non-WG forming BOF - in a nutshell what is the problem?
(not even the scope yet)

If it is performance optimization over interconnected links with different
delay/loss characteristics and with or without host collaboration and/or
access I am sorry to tell you that there has been a lot of work done in
industry and academia addressing this.

Whatever you do you will hit t the fact that a lot of the network coding or
FEC you need is protected. Will LOOPS develop at new code? If yes then
LOOPS is NWCWG the IETF version of our RG. You will also hit the tunneling
limitations and of course proprietary PEPs (the open source ones are okish
though) and the usual issues of congestion losses vs. random losses and we
have looked at this for ages now it seems and we have gotten good at it.

I disagree with the assessment that “long losses” are due to congestion on
your path, they may be due to issues on another path and in wireless some
kind of deep fade that the lower layers did not catch - beware that “delay”
is also a sign of loss.

I agree with some of the discussions today
<http://airmail.calendar/2019-07-23%2012:00:00%20GMT-4>: what layer are we
talking about here? L3?  But some of the diagrams showed OLT and below  L3.
 “Solving" problems at L3 without looking at what is either below or above
is I think something the proposing team should look at. It has helped me in
the past.

A combination of the work we did in NWCRG and what has also been addressed
in ICNRG and ICCRG (and others) is what LOOPS needs. DO we need a new WG to
get that work out of IRTF?

But if the goal of this group is to make progress there are other issues to
address. Mine would be how to implement L3 solutions in an L4 encrypted
world. And others.

Just to say you are there to improve Internet performance of course is
good: we are all here to do that.

mjm

Marie-José Montpetit, Ph.D.
Research Affiliate, MIT Media Laboratory
[e-mail]
[e-mail]