FYI/A: Bufferbloat 101

X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
I can install netperf through Entware on my Asus RT-AX82U as well.

Code: 
:/opt/home/OpenWrtScripts-master# ./betterspeedtest.sh
2021-11-16 21:27:42 Testing against netperf.bufferbloat.net (ipv4) with 5 simultaneous sessions while pinging gstatic.com (60 seconds in each direction)
.............................................................
. Download: 302.40 Mbps
  Latency: (in msec, 55 pings, 0.00% packet loss)
      Min: 176.000
    10pct: 176.000
   Median: 176.000
      Avg: 177.073
    90pct: 179.000
      Max: 194.000
..............................................................
   Upload: 290.82 Mbps
  Latency: (in msec, 62 pings, 0.00% packet loss)
      Min: 176.000
    10pct: 176.000
   Median: 180.000
      Avg: 185.935
    90pct: 195.000
      Max: 264.000
 
hwzlite

hwzlite

Master Member
Joined
Jan 27, 2007
Messages
3,065
Reaction score
3,204
xiaofan said:
SQM is disabled for download (set to 0 in GUI). Queue disipline is set to fq_codel, script is set to piece_of_cake.qos.
Click to expand...

How about Queue disipline is set to cake, script is set to piece_of_cake.qos , any improvement?
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
hwzlite said:
How about Queue disipline is set to cake, script is set to piece_of_cake.qos , any improvement?
Click to expand...

Actually because of the test server problems with DSLreports and Waveform.com (apparently they are not good enough to test with 1Gbps connection, at least for Singapore users), all my wired tests on the Raspberry Pi 400 running OpenWRT does not make too much sense. Sometimes it may lead to the conclusion that SQM-QoS is useless. Sometimes I got A+ without SQM-QoS and only A with SQM-QoS.

Without sorting out the test server issues, I am not going to test anything further with the OpenWRT SQM-QoS settings, because the results are inconsistent.

I will still keep my current settings.
1) No QoS on the Asus RT-AX82U
2) Keep the same SQM-QoS settings to limit the upload speed to 500Mbps.

In the end, my current view is still that QoS or SQM-QoS is useless for 1Gbps connection, at least for my setup (both RT-AX82U and Raspberry Pi 400 perform good enough) and use cases (no gaming, mainly wireless, never really saturate the connection other than running SpeedTest).

But if you encounter issues in real world use cases, then you can try to enable the QoS (eg: Asus Adaptive QoS) or SQM-QoS (eg: OpenWRT, pfSense, etc) on the router to see if that helps.

Just remember that there is limitation with the test servers and they may not be the real judgement for the real bufferbloat situation for your use case.
 
Last edited:
firesong

firesong

Supremacy Member
Deluxe Member
Joined
Jan 17, 2001
Messages
8,811
Reaction score
4,828
I haven't bothered to tune my connection, though I left the limiters configured at 950Mbps up/down. Am not sure if it actually does anything though.
 
hwzlite

hwzlite

Master Member
Joined
Jan 27, 2007
Messages
3,065
Reaction score
3,204
xiaofan said:
But if you encounter issues in real world use cases, then you can try to enable the QoS (eg: Asus Adaptive QoS) or SQM-QoS (eg: OpenWRT, pfSense, etc) on the router to see if that helps.

Another
Click to expand...

Another good way to monitor in SQM level is using the "tc -d qdisc" command.
Checkout this great post on interpreting its result,

Here's mine stats: on my puny DD-WRT (D-Link DIR-868L) router: :giggle:

Code: 
qdisc cake 8001: root refcnt 2 bandwidth 475000Kbit diffserv4 dual-srchost nat nowash ack-filter split-gso rtt 100.0ms raw overhead 0 no-sce
 Sent 72621043268 bytes 167416428 pkt (dropped 5506, overlimits 29614729 requeues 0)
 backlog 0b 0p requeues 0
 memory used: 237312b of 15180Kb
 capacity estimate: 475000Kbit
 min/max network layer size:           39 /    1514
 min/max overhead-adjusted size:       39 /    1514
 average network hdr offset:           14

                   Bulk  Best Effort        Video        Voice
  thresh      29687Kbit   475000Kbit   237500Kbit   118750Kbit
  target          5.0ms        5.0ms        5.0ms        5.0ms
  interval      100.0ms      100.0ms      100.0ms      100.0ms
  pk_delay         17us        2.4ms         21us         35us
  av_delay          5us        291us          9us         12us
  sp_delay          5us          6us          5us          6us
  backlog            0b           0b           0b           0b
  pkts              213    166112666          777      1308278
  bytes           11502  72395778322        41958    231029191
  way_inds            0     26676932            1        21743
  way_miss          115      1307322          333         2643
  way_cols            0           15            0            0
  marks               0            3            0            0
  drops               0         4719            0            0
  ack_drop            0          787            0            0
  sp_flows            1            4            1            1
  bk_flows            0            1            0            0
  un_flows            0            0            0            0
  max_len            54         1514           54         1166
  quantum           905         1514         1514         1514
 
Last edited:
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
hwzlite said:
Another good way to monitor in SQM level is using the "tc -d qdisc" command.
Checkout this great post on interpreting its result,

Here's mine stats: on my puny DD-WRT (D-Link DIR-868L) router: :giggle:
Click to expand...

This command does not work as intended on my Asus router. I guess the tc command from Asus Stock FW is not a cut down version. I will try OpenWRT later.

Code: 
:/tmp/home/root# tc -d qdisc
qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth1 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth2 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth3 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth4 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth5 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
qdisc pfifo_fast 0: dev eth6 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1

:/tmp/home/root# tc -s -d qdisc show
qdisc pfifo_fast 0: dev eth0 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 177169591874 bytes 130941933 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0 
qdisc pfifo_fast 0: dev eth1 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 48113 bytes 215 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0 
qdisc pfifo_fast 0: dev eth2 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0 
qdisc pfifo_fast 0: dev eth3 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 2018049515 bytes 6857292 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0 
qdisc pfifo_fast 0: dev eth4 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 88069968408 bytes 117842777 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0 
qdisc pfifo_fast 0: dev eth5 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 438 bytes 5 pkt (dropped 6774566, overlimits 0 requeues 3) 
 backlog 236415b 1000p requeues 3 
qdisc pfifo_fast 0: dev eth6 root refcnt 2 bands 3 priomap  1 2 2 2 1 2 0 0 1 1 1 1 1 1 1 1
 Sent 6433902299 bytes 14886787 pkt (dropped 0, overlimits 0 requeues 0) 
 backlog 0b 0p requeues 0
 
Last edited:
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
Output from my OpenWRT router, the output is rather cryptic...

Code: 
root@OpenWrt:~# tc -d qdisc
qdisc noqueue 0: dev lo root refcnt 2
qdisc mq 0: dev eth0 root
qdisc fq_codel 0: dev eth0 parent :5 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc fq_codel 0: dev eth0 parent :4 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc fq_codel 0: dev eth0 parent :3 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc fq_codel 0: dev eth0 parent :2 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc fq_codel 0: dev eth0 parent :1 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc cake 801a: dev eth1 root refcnt 2 bandwidth 500Mbit besteffort triple-isolate nonat nowash no-ack-filter split-gso rtt 100ms raw overhead 0
qdisc fq_codel 0: dev tun0 root refcnt 2 limit 10240p flows 1024 quantum 1500 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
qdisc noqueue 0: dev br-lan root refcnt 2
qdisc noqueue 0: dev wg_lan root refcnt 2

The following command seems to give more detailed output. Not so sure what the results mean though...
Code: 
root@OpenWrt:~#  tc -s -d qdisc show
qdisc noqueue 0: dev lo root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc mq 0: dev eth0 root
 Sent 218586915899 bytes 183065039 pkt (dropped 14355, overlimits 0 requeues 7909083)
 backlog 0b 0p requeues 7909083
qdisc fq_codel 0: dev eth0 parent :5 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 39315841820 bytes 34052426 pkt (dropped 6915, overlimits 0 requeues 298461)
 backlog 0b 0p requeues 298461
  maxpacket 1514 drop_overlimit 0 new_flow_count 98802 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :4 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 42314723891 bytes 35718700 pkt (dropped 7438, overlimits 0 requeues 329200)
 backlog 0b 0p requeues 329200
  maxpacket 1514 drop_overlimit 0 new_flow_count 226456 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :3 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 41360989011 bytes 34493688 pkt (dropped 0, overlimits 0 requeues 480843)
 backlog 0b 0p requeues 480843
  maxpacket 1514 drop_overlimit 0 new_flow_count 239994 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :2 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 51798252873 bytes 41531515 pkt (dropped 0, overlimits 0 requeues 6749862)
 backlog 0b 0p requeues 6749862
  maxpacket 1514 drop_overlimit 0 new_flow_count 586117 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :1 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 43797108304 bytes 37268710 pkt (dropped 2, overlimits 0 requeues 50717)
 backlog 0b 0p requeues 50717
  maxpacket 1514 drop_overlimit 0 new_flow_count 50567 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc cake 801a: dev eth1 root refcnt 2 bandwidth 500Mbit besteffort triple-isolate nonat nowash no-ack-filter split-gso rtt 100ms raw overhead 0
 Sent 363039059 bytes 2305798 pkt (dropped 0, overlimits 92504 requeues 1)
 backlog 0b 0p requeues 1
 memory used: 110080b of 15140Kb
 capacity estimate: 500Mbit
 min/max network layer size:           42 /    1514
 min/max overhead-adjusted size:       42 /    1514
 average network hdr offset:           14

                  Tin 0
  thresh        500Mbit
  target            5ms
  interval        100ms
  pk_delay         14us
  av_delay          2us
  sp_delay          1us
  backlog            0b
  pkts          2305798
  bytes       363039059
  way_inds        24490
  way_miss        46711
  way_cols            0
  drops               0
  marks               0
  ack_drop            0
  sp_flows            1
  bk_flows            1
  un_flows            0
  max_len         12484
  quantum          1514

qdisc fq_codel 0: dev tun0 root refcnt 2 limit 10240p flows 1024 quantum 1500 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 264363275 bytes 256867 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
  maxpacket 1401 drop_overlimit 0 new_flow_count 8254 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc noqueue 0: dev br-lan root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc noqueue 0: dev wg_lan root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
Change to Cake and piece_of_cake.qos.

Code: 
root@OpenWrt:~#  tc -s -d qdisc show
qdisc noqueue 0: dev lo root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc mq 0: dev eth0 root
 Sent 218634977183 bytes 183103830 pkt (dropped 14355, overlimits 0 requeues 7909142)
 backlog 0b 0p requeues 7909142
qdisc fq_codel 0: dev eth0 parent :5 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 39316023119 bytes 34052841 pkt (dropped 6915, overlimits 0 requeues 298461)
 backlog 0b 0p requeues 298461
  maxpacket 1514 drop_overlimit 0 new_flow_count 98821 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :4 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 42315189866 bytes 35719349 pkt (dropped 7438, overlimits 0 requeues 329202)
 backlog 0b 0p requeues 329202
  maxpacket 1514 drop_overlimit 0 new_flow_count 226463 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :3 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 41408087034 bytes 34530575 pkt (dropped 0, overlimits 0 requeues 480900)
 backlog 0b 0p requeues 480900
  maxpacket 1514 drop_overlimit 0 new_flow_count 240048 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :2 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 51798359394 bytes 41531789 pkt (dropped 0, overlimits 0 requeues 6749862)
 backlog 0b 0p requeues 6749862
  maxpacket 1514 drop_overlimit 0 new_flow_count 586117 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc fq_codel 0: dev eth0 parent :1 limit 10240p flows 1024 quantum 1514 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 43797317770 bytes 37269276 pkt (dropped 2, overlimits 0 requeues 50717)
 backlog 0b 0p requeues 50717
  maxpacket 1514 drop_overlimit 0 new_flow_count 50570 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc cake 801d: dev eth1 root refcnt 2 bandwidth 500Mbit besteffort triple-isolate nonat nowash no-ack-filter split-gso rtt 100ms raw overhead 0
 Sent 34028 bytes 304 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
 memory used: 2752b of 15140Kb
 capacity estimate: 500Mbit
 min/max network layer size:           42 /     488
 min/max overhead-adjusted size:       42 /     488
 average network hdr offset:           10

                  Tin 0
  thresh        500Mbit
  target            5ms
  interval        100ms
  pk_delay         19us
  av_delay          5us
  sp_delay          3us
  backlog            0b
  pkts              304
  bytes           34028
  way_inds            0
  way_miss           58
  way_cols            0
  drops               0
  marks               0
  ack_drop            0
  sp_flows            1
  bk_flows            0
  un_flows            0
  max_len           488
  quantum          1514

qdisc fq_codel 0: dev tun0 root refcnt 2 limit 10240p flows 1024 quantum 1500 target 5ms interval 100ms memory_limit 4Mb ecn drop_batch 64
 Sent 264363275 bytes 256867 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
  maxpacket 1401 drop_overlimit 0 new_flow_count 8254 ecn_mark 0
  new_flows_len 0 old_flows_len 0
qdisc noqueue 0: dev br-lan root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
qdisc noqueue 0: dev wg_lan root refcnt 2
 Sent 0 bytes 0 pkt (dropped 0, overlimits 0 requeues 0)
 backlog 0b 0p requeues 0
root@OpenWrt:~#
 
hwzlite

hwzlite

Master Member
Joined
Jan 27, 2007
Messages
3,065
Reaction score
3,204
xiaofan said:
Output from my OpenWRT router, the output is rather cryptic...
Click to expand...


Oh my bad, "tc -s -d qdisc show" is the right command.

Code: 
qdisc cake 8001: dev vlan2 root refcnt 2 bandwidth 475000Kbit diffserv4 dual-srchost nat nowash ack-filter split-gso rtt 100.0ms raw overhead 0 no-sce
 Sent 81764854691 bytes 179219095 pkt (dropped 6504, overlimits 34556624 requeues 0)
 backlog 0b 0p requeues 0
 memory used: 237312b of 15180Kb
 capacity estimate: 475000Kbit
 min/max network layer size:           39 /    1514
 min/max overhead-adjusted size:       39 /    1514
 average network hdr offset:           14

                   Bulk  Best Effort        Video        Voice
  thresh      29687Kbit   475000Kbit   237500Kbit   118750Kbit
  target          5.0ms        5.0ms        5.0ms        5.0ms
  interval      100.0ms      100.0ms      100.0ms      100.0ms
  pk_delay         17us         29us         19us         20us
  av_delay          5us         14us          9us         11us
  sp_delay          5us          8us          5us          5us
  backlog            0b           0b           0b           0b
  pkts              220    177858713          821      1365845
  bytes           11880  81530743244        44334    240761955
  way_inds            0     27316049            1        23462
  way_miss          120      1395365          355         2743
  way_cols            0           15            0            0
  marks               0            3            0            0
  drops               0         5353            0            0
  ack_drop            0         1151            0            0
  sp_flows            1            4            1            1
  bk_flows            0            1            0            0
  un_flows            0            0            0            0
  max_len            54         1514           54         1166
  quantum           905         1514         1514         1514

For my case, i did not use the default Cake configuration from DD-WRT GUI.

I manually configured via CLI with this tweaked setting:
Code: 
tc qdisc add dev vlan2 root cake ack-filter bandwidth 475000kbit diffserv4 nat dual-srchost

Basically diffserv4 setting provides 4 bins of priority (Bulk ,Best Effort , Video , Voice) .
i also manually DSCP tagged at router iptable CLI on my torrenting & Tor IP host "192.168.1.8" to lowest "Bulk" bin, and RipeAtlasProbe "192.168.1.10" at highest "Video" bin .

Code: 
iptables -t mangle -A PREROUTING  -p udp --src 192.168.1.8 -j DSCP --set-dscp-class CS1
iptables -t mangle -A PREROUTING  -p tcp --src 192.168.1.8 -j DSCP --set-dscp-class CS1
iptables -t mangle -A PREROUTING  -p udp --src 192.168.1.10 -j DSCP --set-dscp-class CS3
iptables -t mangle -A PREROUTING  -p tcp --src 192.168.1.10 -j DSCP --set-dscp-class CS3

Check out https://man7.org/linux/man-pages/man8/tc-cake.8.html for other tweak parameters.

DSCP references for keeping myself sane :grin::
MS Teams
Zoom
 
Last edited:
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
hwzlite said:
tc -s -d qdisc show
Click to expand...

Yes I have provided the output from my Asus RT-AX82U (without any QoS) and Raspberry Pi 400 running OpenWRT (with SQM QoS, with fq_codel or with cake.

It seems to me the results do not tell much.

Guess it is not so easy to test bufferbloat with an 1Gbps connection since servers from DSLreports and Waveform.com do not seem to be good enough. And if we set up a netperf server on the local PC the results may be too good to really show any differences.

On the other hand, probably results from DSLreports and Waveform.com can still be of some reference values for both wired and wireless connections. And the scripts/tools mentioned below are also useful.
https://www.bufferbloat.net/projects/bloat/wiki/Tests_for_Bufferbloat/
 
Last edited:
firesong

firesong

Supremacy Member
Deluxe Member
Joined
Jan 17, 2001
Messages
8,811
Reaction score
4,828
Untuned, but with limiters in place for 950Mbps, on wireless. (So too many variables).

Did a network sweep and realised a neighbour has a new strong wireless connection that's set too high. There's definitely a lot of network congestion going on, even at 5Ghz (suspect it's the new neighbour that recently moved in below my unit)

https://www.waveform.com/tools/bufferbloat?test-id=f7f3f71f-fa04-454f-bb7c-343b82ffe09c
What I find interesting are these "stray pings" that are probably the real result of bufferbloat, yet happen the least frequently. I would be more concerned if I get a lot of dots at those ends, but I guess I could live with 1 or 2. And no free-fragging for me.
Bufferbloat_Test_by_Waveform.png
 
Last edited:
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
firesong said:
Untuned, but with limiters in place for 950Mbps, on wireless. (So too many variables).

Did a network sweep and realised a neighbour has a new strong wireless connection that's set too high. There's definitely a lot of network congestion going on, even at 5Ghz (suspect it's the new neighbour that recently moved in below my unit)

https://www.waveform.com/tools/bufferbloat?test-id=f7f3f71f-fa04-454f-bb7c-343b82ffe09c
Click to expand...

You may want to try to limit the speed to 500Mbps or lower for wireless testing and probably you will see an improvement. Setting it to 950Mbps will have no effect on wireless and actually not really that useful even for wired. It is similar to disable the SQM QoS.
 
firesong

firesong

Supremacy Member
Deluxe Member
Joined
Jan 17, 2001
Messages
8,811
Reaction score
4,828
xiaofan said:
You may want to try to limit the speed to 500Mbps or lower for wireless testing and probably you will see an improvement. Setting it to 950Mbps will have no effect on wireless and actually not really that useful even for wired. It is similar to disable the SQM QoS.
Click to expand...
Heh. My first test results doing this.

https://www.waveform.com/tools/bufferbloat?test-id=a141b701-8443-47e7-8921-31ba1670a39a
Not quite the desired outcome in spite of the A+ rating. ;)

The second isn't too far off from not having it at 500/500:
https://www.waveform.com/tools/bufferbloat?test-id=9eb1aa69-25d9-494e-b68e-fbb5bce30aea
The third suggests it's still wireless latency:
https://www.waveform.com/tools/bufferbloat?test-id=7de56375-87be-41a7-a0ad-ae42ea2854d1
Edit: Checked. Limiters are working (cos I'm also speedtesting on the box).
Will just leave it be at 950. I have wired clients, including my NAS, that benefit from the higher throughputs. And I will rue the new neighbour's disruptive wifi. :/

Code: 
fs@Khadas:~$ speedtest

   Speedtest by Ookla

     Server: PT FirstMedia - Singapore (id = 7556)
        ISP: Singtel Fiber
    Latency:     1.93 ms   (0.07 ms jitter)
   Download:   482.89 Mbps (data used: 217.3 MB )                               
     Upload:   482.62 Mbps (data used: 530.5 MB )                               
Packet Loss: Not available.
 Result URL: https://www.speedtest.net/result/c/5fad08ab-df82-4ce3-b3c2-220a66e62032

fs@Khadas:~$ speedtest

   Speedtest by Ookla

     Server: Singtel - Singapore (id = 25960)
        ISP: Singtel Fiber
    Latency:     1.66 ms   (0.11 ms jitter)
   Download:   948.20 Mbps (data used: 699.4 MB )                               
     Upload:   944.72 Mbps (data used: 1.1 GB )                               
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/b9f137be-5d09-4514-b05d-c3b156da1256
 
Last edited:
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
firesong said:
Edit: Checked. Limiters are working (cos I'm also speedtesting on the box).
Will just leave it be at 950. I have wired clients, including my NAS, that benefit from the higher throughputs. And I will rue the new neighbour's disruptive wifi. :/

Code: 
fs@Khadas:~$ speedtest
fs@Khadas:~$ speedtest

   Speedtest by Ookla

     Server: Singtel - Singapore (id = 25960)
        ISP: Singtel Fiber
    Latency:     1.66 ms   (0.11 ms jitter)
   Download:   948.20 Mbps (data used: 699.4 MB )                             
     Upload:   944.72 Mbps (data used: 1.1 GB )                             
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/b9f137be-5d09-4514-b05d-c3b156da1256
Click to expand...

Actually I can not tell whether the limiter is really working or not because that is the typical wired speed you will get for 1Gbps plan, similar to my wired speed without any SQM-QoS settings. You may have to see what is the line speed without the SQM QoS settings as well.

I will say setting it to 900Mbps (at least for the upload) and the test will make the results clearer.
 
firesong

firesong

Supremacy Member
Deluxe Member
Joined
Jan 17, 2001
Messages
8,811
Reaction score
4,828
xiaofan said:
Actually I can not tell whether the limiter is really working or not because that is the typical wired speed you will get for 1Gbps plan, similar to my wired speed without any SQM-QoS settings. You may have to see what is the line speed without the SQM QoS settings as well.

I will say setting it to 900Mbps (at least for the upload) and the test will make the results clearer.
Click to expand...
Oh, I wasn't clear.

The 482MB speed was with a 500Mbps limiter - which resulted in the 3 waveform tests I'd linked. The first was a hilarious result.

The 2nd one was with a 980Mbps limiter. Was just copy/pasted to show a contrast.
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
firesong said:
Oh, I wasn't clear.
The 482MB speed was with a 500Mbps limiter - which resulted in the 3 waveform tests I'd linked. The first was a hilarious result.
The 2nd one was with a 980Mbps limiter. Was just copy/pasted to show a contrast.
Click to expand...

I see. Thanks.

So your 500Mbps limiter is working fine but indeed your first A+ result does not make much sense because of the high base line ping value.

Your second and 3rd score of B is more likely due to the latency using wireless.

All in all I think results using wireless tend to be inconsistent and often it kind of produces results showing that SQM QoS does not make much or any differences.
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
miloaisdino said:
haha my mikrotik is connected to the tplink switch then to the ONT. had to put a cap on the upload speed as the newer M1 ONTs issued seem to be lower spec (with cheap plastiky feel) else would experience severe bufferbloat on upload again. without the rate limit on my switch the max upload speed i could record on speedtest.net was 850mbps anyway so not much of a performance penalty
:)
Click to expand...

Interesting that you need to use the TL-SG105E again with M1 ONT.

What is the ONT given by M1? Previously I know it is good Huawei ONT.
https://www.m1.com.sg/-/media/Files/Support/FAQ/Fibre Broadband/ONT_Guide.pdf
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
Tested under Ubuntu Linux 20.04 on my new Acer Swift 3 laptop with Intel AX201 adapter, Asus RT-AX82U router without any QoS settings. Singtel 1Gbps plan with ZTE ONT (which may have bufferbloat issues above 800Mbps for upload).

OOkla speedtest using CLI version.
Code: 
$ speedtest -s 25960
   Speedtest by Ookla
     Server: Singtel - Singapore (id = 25960)
        ISP: Singtel Fiber
    Latency:     2.35 ms   (0.34 ms jitter)
   Download:   599.04 Mbps (data used: 733.6 MB )                               
     Upload:   707.35 Mbps (data used: 1.2 GB )                               
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/6b24c952-5a31-4b6a-acf1-f6ba825fcd6b

1) A+ score using wireless, download 838.7Mbps, upload 395.3Mbps, the base line ping is higher and that is why A+ score
https://www.waveform.com/tools/bufferbloat?test-id=fa84e0e3-c1be-417a-bd65-682a5cd9fc6a
2) A score using wireless, download 761.2Mbps, upload 261.8Mbps, base line ping is low at 4ms.
https://www.waveform.com/tools/bufferbloat?test-id=a9b2b56e-b2d9-49eb-8d87-d769dd10727a
 
X

xiaofan

High Supremacy Member
HWZ Insider
Joined
Sep 16, 2018
Messages
34,820
Reaction score
11,557
Wired tests under Ubuntu Linux 20.04 on my new Acer Swift 3 laptop with Ugreen USB 3 to 2.5G Ethernet adapter, Asus RT-AX82U router without any QoS settings. Singtel 1Gbps plan with ZTE ONT (which may have bufferbloat issues above 800Mbps+ for upload).

OOkla speedtest using CLI version.
Code: 
speedtest -s 25960

   Speedtest by Ookla

     Server: Singtel - Singapore (id = 25960)
        ISP: Singtel Fiber
    Latency:     2.47 ms   (0.23 ms jitter)
   Download:   937.44 Mbps (data used: 796.6 MB )                               
     Upload:   945.65 Mbps (data used: 1.1 GB )                               
Packet Loss:     0.0%
 Result URL: https://www.speedtest.net/result/c/8c2fdd74-c1b6-4fd1-9a7b-a1d8044eeb4e

1) bufferbloat score of A, download 874.8Mbps, upload 774.3Mbps
https://www.waveform.com/tools/bufferbloat?test-id=b4859e02-228c-4b0b-b9c3-4c27846124b7
2) bufferbloat score of A, download 884.0Mbps, upload 816.6Mbps
https://www.waveform.com/tools/bufferbloat?test-id=eed981db-d3ca-4463-947c-91f7ab0e1672
 
You must log in or register to reply here.
Important Forum Advisory Note
This forum is moderated by volunteer moderators who will react only to members' feedback on posts. Moderators are not employees or representatives of HWZ Forums. Forum members and moderators are responsible for their own posts. Please refer to our Community Guidelines and Standards and Terms and Conditions for more information.
Top