[Ecn-sane] [iccrg] [tsvwg] IETF104 ICCRG TCP Prague presentation

[Dave Taht]

Several things...

I find discussions of ECN in general very stressful, as I am firmly on
the ecn-sane "yellow" team. ECN scares me on the general internet.
More than anything else, I'd like to get a big DDOS protection
provider and university to tackle all my outstanding issues with using
ECN in a general deployment - ranging from what happens with tcp syn
ecn enabled attacks, to ddos protection from udp based packet floods,
as I outlined in the ecn-sane project's charter:
https://www.bufferbloat.net/projects/ecn-sane/wiki/ before I can
endorse this work.

I have friends that regularly deal with 2TByte/sec DDOS attacks and
they have worse blood pressure problems than I do.

I've long had no problem with ecn in specialized deployments, under
rigidly controlled conditions. The SCE work, in particular, looks to
be useful for technologies like nvme over tcp and roce and stuff like
that, and I look forward to results there.

One of the proudest things I have ever had going in my life is that
the backhauls and groundstations for the arkyd-6 spacecraft, used
fq_codel with ECN enabled... running lede (openwrt). And though I
can't talk about that project much, and probably shouldn't have
mentioned even that much here  - I tend to feel that any non-backward
compatible change to ecn will muck with *my* bird in a 25 year orbit,
and I kind of take it a bit personally any naive interference with
that contribution to the US space program.

I am unsubscribing from the ietf lists as of now, so I can focus on
getting testbeds and analytical tools up, as well as finishing up my
work on the ipv4 unicast extensions project, which is entirely
separate, that I wish I could get more folk to pay attention to.

If there is a productive conversation of interest to me, here, using
debating techniques I can live with (
https://www.bufferbloat.net/projects/ecn-sane/wiki/rules/ ) please
either send a link or cc ecn-sane at lists.bufferbloat.net.

Thank you.

Dave