Cake - FQ_codel the next generation
 help / color / mirror / Atom feed
From: Adrian Popescu <adriannnpopescu@gmail.com>
To: Dave Taht <dave.taht@gmail.com>
Cc: cake@lists.bufferbloat.net
Subject: Re: [Cake] 450 up stream test
Date: Thu, 23 Apr 2015 19:35:22 +0300	[thread overview]
Message-ID: <CAF3M4P2=2LWKgnMbGPMfMMAoRUQnjd-CPP-3ANzqXxYf=SbJBg@mail.gmail.com> (raw)
In-Reply-To: <CAA93jw74TKyi4WMfDLrrVG9VbDYZ+RnGG+Dv=5sp38QitkfRpw@mail.gmail.com>

I've also observed higher packet delay on the second class with a
single stream with a rate limiter. Setting a rate limit of 50mbps
seems to be inducing latency as that stream goes up towards the limit.
I haven't checked if there were other competing streams which were
actually seeing that latency.

On Wed, Apr 22, 2015 at 11:34 AM, Dave Taht <dave.taht@gmail.com> wrote:
> I was finally able to induce "way" collisions and codel was not able
> to successfully control the depth of the queue in this test of 9
> copies of the 50 up test....
>
> root@nuc-client:~/git/iproute2-cake# tc -s qdisc show dev eth0
> qdisc cake 8003: root refcnt 2 bandwidth 920Mbit diffserv4 flows
>  Sent 13826214606 bytes 10775480 pkt (dropped 187301, overlimits
> 7841897 requeues 79)
>  backlog 1732937b 769p requeues 79
>            Class 0     Class 1     Class 2     Class 3
>   rate       920Mbit  862500Kbit     690Mbit     230Mbit
>   target       5.0ms       5.0ms       5.0ms       5.0ms
> interval     105.0ms     105.0ms     105.0ms     105.0ms
> Pk delay         0us      22.0ms         0us        19us
> Av delay         0us      12.0ms         0us         1us
> Sp delay         0us       3.8ms         0us         1us
>   pkts        305686     4248957           0      520011
> way inds           0      504555           0           0
> way miss           3      165303           0           7
> way cols           0        4656           0           0
>   bytes   1040146583 11540379317           0  1730046460
>   drops           38      187176           0          87
>   marks            0           0           0           0
>
> We dont  have a good way to plot a test this extreme, and offloads do
> hurt a lot. I turned off offloads to see what would happen:
>
> root@nuc-client:~/git/iproute2-cake# tc -s qdisc show dev eth0
> qdisc cake 8004: root refcnt 2 bandwidth 920Mbit diffserv4 flows
>  Sent 3147074201 bytes 2948700 pkt (dropped 115035, overlimits 4207356
> requeues 417)
>  backlog 1592366b 1159p requeues 417
>            Class 0     Class 1     Class 2     Class 3
>   rate       920Mbit  862500Kbit     690Mbit     230Mbit
>   target       5.0ms       5.0ms       5.0ms       5.0ms
> interval     105.0ms     105.0ms     105.0ms     105.0ms
> Pk delay         0us      26.2ms         0us         9us
> Av delay         0us      12.2ms         0us         0us
> Sp delay         0us       4.7ms         0us         0us
>   pkts             0     3064884           0          10
> way inds           0      456896           0           0
> way miss           0      106191           0           2
> way cols           0        1085           0           0
>   bytes            0  3322825235           0        1186
>   drops            0      115035           0           0
>   marks            0           0           0           0
>
>
> --
> Dave Täht
> Open Networking needs **Open Source Hardware**
>
> https://plus.google.com/u/0/+EricRaymond/posts/JqxCe2pFr67
> _______________________________________________
> Cake mailing list
> Cake@lists.bufferbloat.net
> https://lists.bufferbloat.net/listinfo/cake

      reply	other threads:[~2015-04-23 16:35 UTC|newest]

Thread overview: 2+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-04-22  8:34 Dave Taht
2015-04-23 16:35 ` Adrian Popescu [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

  List information: https://lists.bufferbloat.net/postorius/lists/cake.lists.bufferbloat.net/

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to='CAF3M4P2=2LWKgnMbGPMfMMAoRUQnjd-CPP-3ANzqXxYf=SbJBg@mail.gmail.com' \
    --to=adriannnpopescu@gmail.com \
    --cc=cake@lists.bufferbloat.net \
    --cc=dave.taht@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox