testing classification on wireless

Tue Apr 3 16:48:31 EDT 2012

The newest netperf from svn has the ability to set diffserv values both
locally and remotely. This makes it really easy to exercise 802.11e queues,
in either direction, in multiple ways, under various workloads. (hat off to
rick jones for the work)

This is also the very first time I've seen the ath9k driver statistics look
sane to me (haven't checked in months tho)
(hats off to many folk for that)

ath9k stats: http://pastebin.com/2kfJkQ4n

simple example of use

m at cruithne:~/classes$ more classtest
#!/bin/sh

BK="CS1"
BE="CS0"
VI="CS5"
VO="EF"
netperf -l 60 -Y $BK,$BK -H 172.29.1.1 >> $BK.log &
netperf -l 60 -Y $VI,$VI -H 172.29.1.1 >> $VI.log &
netperf -l 60 -Y $BE,$BE -H 172.29.1.1 >> $BE.log &
netperf -l 60 -Y $VO,$VO -H 172.29.1.1 >> $VO.log &

m at cruithne:~/classes$ more CS0.log
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
172.29.1.1 () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 87380  65536  65536    60.04      34.68

m at cruithne:~/classes$ more CS1.log
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
172.29.1.1 () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 87380  65536  65536    60.07       2.62

m at cruithne:~/classes$ more CS5.log
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
172.29.1.1 () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 87380  65536  65536    60.01      16.90

m at cruithne:~/classes$ more EF.log
MIGRATED TCP STREAM TEST from 0.0.0.0 (0.0.0.0) port 0 AF_INET to
172.29.1.1 () port 0 AF_INET
Recv   Send    Send
Socket Socket  Message  Elapsed
Size   Size    Size     Time     Throughput
bytes  bytes   bytes    secs.    10^6bits/sec

 87380  65536  65536    60.02      16.30

(this was through sfqred on both sides, which could use some tuning for
each queue, and I need to fiddle with tcp_rr, but I have a few other fish
to fry this week)

m at cruithne:~/classes$ tc -s qdisc show dev wlan0
qdisc mq 1: root
 Sent 5352889590 bytes 4677961 pkt (dropped 66, overlimits 1278 requeues
173)
 backlog 0b 0p requeues 173

qdisc sfq 10: parent 1:1 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 6 prob_drop 0
 forced_mark 0 forced_mark_head 13 forced_drop 0
 Sent 175003610 bytes 190526 pkt (dropped 2, overlimits 19 requeues 4)
 rate 4528bit 1pps backlog 0b 0p requeues 4

qdisc sfq 20: parent 1:2 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 173 prob_drop 0
 forced_mark 0 forced_mark_head 54 forced_drop 0
 Sent 1774758920 bytes 1176424 pkt (dropped 9, overlimits 227 requeues 68)
 rate 0bit 0pps backlog 0b 0p requeues 68

qdisc sfq 30: parent 1:3 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 3 prob_mark_head 317 prob_drop 8
 forced_mark 6 forced_mark_head 692 forced_drop 2
 Sent 2521435696 bytes 2727177 pkt (dropped 55, overlimits 1028 requeues
55)
 rate 1280bit 1pps backlog 0b 0p requeues 55

qdisc sfq 40: parent 1:4 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 4 prob_drop 0
 forced_mark 0 forced_mark_head 0 forced_drop 0
 Sent 881691364 bytes 583834 pkt (dropped 0, overlimits 4 requeues 46)
 rate 0bit 0pps backlog 0b 0p requeues 46

root at europa:~# tc -s qdisc show dev sw10

qdisc mq 1: root
 Sent 70630412 bytes 1003868 pkt (dropped 290, overlimits 0 requeues 3283)
 backlog 0b 0p requeues 3283

qdisc sfq 10: parent 1:1 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 0 prob_drop 0
 forced_mark 0 forced_mark_head 0 forced_drop 0
 Sent 3126011 bytes 45972 pkt (dropped 0, overlimits 0 requeues 2749)
 rate 1200bit 1pps backlog 0b 0p requeues 2749

qdisc sfq 20: parent 1:2 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 0 prob_drop 0
 forced_mark 0 forced_mark_head 0 forced_drop 0
 Sent 35166768 bytes 532192 pkt (dropped 86, overlimits 0 requeues 340)
 rate 0bit 0pps backlog 0b 0p requeues 340

qdisc sfq 30: parent 1:3 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 0 prob_drop 0
 forced_mark 0 forced_mark_head 0 forced_drop 0
 Sent 13624933 bytes 142800 pkt (dropped 87, overlimits 0 requeues 106)
 rate 10736bit 5pps backlog 0b 0p requeues 106

qdisc sfq 40: parent 1:4 limit 200p quantum 3028b depth 24 headdrop divisor
16384 perturb 600sec
 ewma 3 min 4500b max 18000b probability 0.2 ecn
 prob_mark 0 prob_mark_head 0 prob_drop 0
 forced_mark 0 forced_mark_head 0 forced_drop 0
 Sent 18712700 bytes 282904 pkt (dropped 117, overlimits 0 requeues 88)
 rate 0bit 0pps backlog 0b 0p requeues 88

-- 
Dave Täht
SKYPE: davetaht
US Tel: 1-239-829-5608
http://www.bufferbloat.net
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.bufferbloat.net/pipermail/bloat-devel/attachments/20120403/50f3210d/attachment-0002.html>