On 02/28/2016 04:21, Dave Täht wrote:
On 2/26/16 6:17 AM, Rasool Al-Saadi wrote:
[..]
Technical report:
http://caia.swin.edu.au/reports/160226A/CAIA-TR-160226A.pdf
In browsing this it appears that shaped rates were tested only (?).
Yes.
I am
curious what native performance (10,100,1gbit) looked like. I think that
freebsd lacks a BQL-like mechanism to control the driver queues, and on
the other hand freebsd did not go as nuts with offloads as linux did. Is
this code generally applicable (to things like pfsense?)
We haven't tested pfSense, but as their website says pfSense 2.2 is based on
FreeBSD 10.1, and our patch applies to stock ipfw and dummynet under FreeBSD
10.1, it may well simply drop-in just fine.
Aside from that, looks pretty good. I am curious also as to what caused
the offset difference in sawtooth pattern between linux and bsd
implementations (like in fig 2) Different initcwnd? ssthresh? don't seem
to be it - linux reno vs bsd reno?
The only change in e.g. Fig.2 was to reboot the bottleneck router between Linux
and FreeBSD -- the source was the same TCP stack in either case. We haven't dug
into the reasons for observed differences yet, although they may be related to
the slightly higher RTTs observed when using FreeBSD + our patched dummynet.
Future work ;)
cheers,
gja
_______________________________________________
aqm mailing list
aqm@ietf.org
https://www.ietf.org/mailman/listinfo/aqm
