* [Cake] cake at 3gbit
@ 2018-05-30 21:32 Dave Taht
2018-05-31 15:40 ` Toke Høiland-Jørgensen
0 siblings, 1 reply; 2+ messages in thread
From: Dave Taht @ 2018-05-30 21:32 UTC (permalink / raw)
To: Cake List; +Cc: Jesper Dangaard Brouer
my old veth code - which simulates 5 hops - can only achieve about
4gbit on the hardware I have with netem on a 20ms delay. There are
undoubtably other bottlenecks elsewhere in the system in this
configuration. :)
I DO have real 10GigE hardware now, but would love to see someone
trying cake at 40GigE+ in the real world now that it allows
superpackets and the api allows for more than 40Gbit to be specified.
regretably, hit a bug, we need to wait for toke's v18 to come out.
At 3gbit, with netem inserting 10ms delay in each direction, it's hard
to complain.
root@spaceheater:~/git/veth# ip netns exec client flent -H 10.10.0.1
-t cake-3gbit rrul_be
Started Flent 1.1.1-git-5daa2b3 using Python 3.5.2.
Starting rrul_be test. Expected run time: 70 seconds.
Data file written to ./rrul_be-2018-05-30T142933.502257.cake-3gbit.flent.gz.
Summary of rrul_be test run 'cake-3gbit' (at 2018-05-30 21:29:33.502257):
avg median # data pts
Ping (ms) ICMP : 20.32 20.30 ms 349
Ping (ms) UDP BE1 : 48.84 20.48 ms 342
Ping (ms) UDP BE2 : 48.84 20.48 ms 342
Ping (ms) UDP BE3 : 48.84 20.48 ms 342
Ping (ms) avg : 41.71 20.43 ms 351
TCP download BE : 704.57 705.27 Mbits/s 300
TCP download BE2 : 700.29 705.23 Mbits/s 300
TCP download BE3 : 703.67 705.24 Mbits/s 299
TCP download BE4 : 700.28 705.28 Mbits/s 300
TCP download avg : 702.20 705.09 Mbits/s 301
TCP download sum : 2808.81 2820.35 Mbits/s 301
TCP totals : 5615.83 5640.44 Mbits/s 301
TCP upload BE : 704.32 705.66 Mbits/s 286
TCP upload BE2 : 703.59 705.46 Mbits/s 286
TCP upload BE3 : 699.39 705.29 Mbits/s 286
TCP upload BE4 : 699.72 705.38 Mbits/s 285
TCP upload avg : 701.75 705.08 Mbits/s 301
TCP upload sum : 2807.02 2820.31 Mbits/s 301
--
Dave Täht
CEO, TekLibre, LLC
http://www.teklibre.com
Tel: 1-669-226-2619
^ permalink raw reply [flat|nested] 2+ messages in thread
* Re: [Cake] cake at 3gbit
2018-05-30 21:32 [Cake] cake at 3gbit Dave Taht
@ 2018-05-31 15:40 ` Toke Høiland-Jørgensen
0 siblings, 0 replies; 2+ messages in thread
From: Toke Høiland-Jørgensen @ 2018-05-31 15:40 UTC (permalink / raw)
To: Dave Taht, Cake List; +Cc: Jesper Dangaard Brouer
Dave Taht <dave.taht@gmail.com> writes:
> I DO have real 10GigE hardware now, but would love to see someone
> trying cake at 40GigE+ in the real world now that it allows
> superpackets and the api allows for more than 40Gbit to be specified.
I have had no problem saturating a 40Gig link with Cake on it with TCP
traffic. Even with the shaper enabled, I got pretty much to line rate,
IIRC (hence the change switching the rate parameter to 64 bits).
I just got some 100gig cards, so maybe there's a chance we can hit a
bottleneck with those. I'll do a quick test run shot once I get them
installed :)
-Toke
^ permalink raw reply [flat|nested] 2+ messages in thread
end of thread, other threads:[~2018-05-31 15:40 UTC | newest]
Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-05-30 21:32 [Cake] cake at 3gbit Dave Taht
2018-05-31 15:40 ` Toke Høiland-Jørgensen
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox