[Ecn-sane] some l4s and sce gsoc work items planned for ns3
dave.taht at gmail.com
Sat May 16 14:16:54 EDT 2020
forwarding this on behalf of tom who is having trouble with his mailer lately.
---------- Forwarded message ---------
From: Tom Henderson <tomhend at uw.edu>
Date: Fri, May 15, 2020 at 11:38 AM
Subject: Resending: Re: tcp prague bug in ns3 also?
To: <dave.taht at gmail.com>
Dave and all,
I discovered a few minutes ago that my outbound email to Gmail.com
addresses has started to experience some delivery issues this morning,
so you may not have received the below, which I'm now resending.
Dave, I see you just ask ecn-sane; I can try to follow up there as well.
On 5/14/20 11:20 PM, Dave Taht wrote:
do you know if tcp prague bug found (no response to loss in the ll
queue) recently in the linux version... is in the ns3 version?
and where's the current branch?
It should not be susceptible to that; it is a completely different
implementation than the Linux Prague.
Current version has been dormant for a few months and is here:
Current status is different from that of Linux. ns-3 version is DCTCP
plus the 'paced chirping' startup that Joakim Misund added. ns-3
version does not have RTT independence, 3168 detection/fallback, or
pacing alignments with Linux for congestion avoidance phase.
We have a GSoC student (Deepak K, cc'ed in this response) who plans to
align the ns-3 model with the Linux code, adding the missing features.
That project is about to start.
We also have another GSoC project that is currently planned to add SCE
TCP extensions and SCE-based AQMs, although we are waiting to see what
tsvwg decides (if anything) this month.
If you or anyone else is interested in tracking these projects, here
are the initial wiki pages:
and we are setting up public chat threads for each project on this
"For a successful technology, reality must take precedence over public
relations, for Mother Nature cannot be fooled" - Richard Feynman
dave at taht.net <Dave Täht> CTO, TekLibre, LLC Tel: 1-831-435-0729
More information about the Ecn-sane