[Cerowrt-devel] Managed to break 802.11n (on a 3800)

[Sebastian Moeller]

Hi Toke,

On Jan 19, 2014, at 21:58 , Toke Høiland-Jørgensen <toke at toke.dk> wrote:

> Sebastian Moeller <moeller0 at gmx.de> writes:
> 
>> I noticed the new "hostnames are mandatory police" kick in:
> 
> Right, botched that; should be fixed properly now.

	That was fast, and:
bash-3.2$ ./netperf-wrapper --list-tests
Available tests:
  cisco_5tcpup               :  RTT Fair Realtime Response Under Load
  cisco_5tcpup_2udpflood     :  Cisco 5TCP up + 2 6Mbit UDP
  cubic_ledbat_1             :  Cubic vs LEDBAT upload streams w/ping
  ledbat_cubic_1             :  Cubic vs LEDBAT upload streams w/ping
  ping                       :  Straight ping test
  reno_cubic_westwood_ledbat :  Realtime Response Under Load
                                (with different congestion control algs)
  reno_cubic_westwood_lp     :  Realtime Response Under Load
                                (with different congestion control algs)
  rrul                       :  Realtime Response Under Load
  rrul46                     :  Realtime Response Under Load - Mixed IPv4/6
  rrul46compete              :  Realtime Response Under Load - Mixed v4/v6 compete
  rrul_be                    :  Realtime Response Under Load - exclusively Best Effort
  rrul_noclassification      :  Realtime Response Under Load - no classification on data flows
  rtt_fair                   :  RTT Fair Realtime Response Under Load
  rtt_fair4be                :  RTT Fair Realtime Response Under Load
  rtt_fair6be                :  RTT Fair Realtime Response Under Load
  tcp_1down                  :  Bidirectional TCP streams w/ping
  tcp_1up                    :  Single TCP upload stream w/ping
  tcp_bidirectional          :  Bidirectional TCP streams w/ping
  tcp_download               :  TCP download stream w/ping
  tcp_upload                 :  TCP upload stream w/ping
  udp_flood                  :  UDP flood w/ping
bash-3.2$ 

Thanks a lot.

> 
>> Wireless is weird...
> 
> Yes, wayy too much black magic involved…

	What I failed to mention, testing against demo (~32ms unloaded ping RTT) from home gives ping RTT around 50ms, from the university 200ms (both with tcp_bidirectional). Ouch, and this with a massive rsync job running at home on a wired machine(over the atlantic, so sharing resources with tcp_bidirectional). 
	Dave, you are right default wireless performance truly is suboptimal…

best
	Sebastian



> 
> -Toke

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/cerowrt-devel/attachments/20140119/0dd240d2/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tcp_bidirectional_EKUT.png
Type: image/png
Size: 47722 bytes
Desc: not available
URL: <https://lists.bufferbloat.net/pipermail/cerowrt-devel/attachments/20140119/0dd240d2/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tcp_bidirectional_HOME2DEMO.png
Type: image/png
Size: 51837 bytes
Desc: not available
URL: <https://lists.bufferbloat.net/pipermail/cerowrt-devel/attachments/20140119/0dd240d2/attachment-0003.png>