From: Stef Coene <stafke@iname.com>
To: lartc@vger.kernel.org
Subject: Re: [LARTC] How to use tc to limit bandwidth of a special IP in LAN
Date: Mon, 18 Jun 2001 05:40:31 +0000 [thread overview]
Message-ID: <marc-lartc-99442832205183@msgid-missing> (raw)
In-Reply-To: <marc-lartc-99233276410622@msgid-missing>
Ramin Alidousti wrote:
>
> Thanks Stef :-)
>
> On Fri, Jun 15, 2001 at 07:57:12AM +0200, Stef Coene wrote:
>
> > - allot 1514 cell 8 maxburst 10 avpkt 1000 : forget these options for
> > now, just set them
>
> That's exactly what I meant.
>
> > - split 1:0 defmap c0 : ? ? ? ? Maybe to split the traffic? I never
> > needed this options, so I don't know what this mean. Just forget this
> > option for now.
>
> You will never need something when you don't know of its existance ;-)
>
> > > A small note on all these options would have saved me a lot of guess-work.
> > Better? ;-)
>
> Yes. Much better :-)
>
> > > But anyway, don't take me wrong. I think that iproute2 is one of the coolest
> > > networking tools I've ever seen.
> > Me to, you can do great things with CBQ but it really sucks and is not
> > accurate enough.
>
> Why do you need accuracy when you're doing "Traffic Engineering" ;-)
I did some tests : I tried to give 2 hosts a bandwidth og 128kbit/s.
Host A needs 75% of this bandwidth and host B 25%. They share the same
128kbit/s so they may borrow unused bandwidth. The only way to do this
is to create a qdisc on a bounded class of 128kbit/s (I tried all other
things and this was the only way to get it working). On this qdisc you
can attach 2 classes for the two hosts with appropriate optiones and it
works perfect. Host A gets 75% of the bandwidth and they can borrow
bandwidth.
BUT, the problem is the bounded class and qdisc. I did some tests and
CBQ is very bad when it has to bound a class, but the problem is that
you need a bounded class with his rate equal to the link-bandwidth. You
can find some nice pictures on my website and they are very clear
(http://users.belgacom.net/staf/qos/tests/cbq/bounded/bounded-accuracy.html):
CBQ is not accurate when bounding a class. So it's working but not
accurate enough. There are some other problems, but this the most
inportant.
So if anyone out there has other results or find a solution, plz let us
know. I did a lot of testing and I'm surre I didn't made a mistake. I
made all the possible combinations of options, and each time it was not
working like I wanted to.
--
Stef
More QOS info : http://users.belgacom.net/staf/
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://ds9a.nl/2.4Routing/
next prev parent reply other threads:[~2001-06-18 5:40 UTC|newest]
Thread overview: 8+ messages / expand[flat|nested] mbox.gz Atom feed top
2001-06-12 7:59 [LARTC] How to use tc to limit bandwidth of a special IP in LAN Chuanbo Xu
2001-06-12 9:39 ` Manfred Bartz
2001-06-14 3:13 ` Bob Puff@NLE
2001-06-15 5:57 ` Stef Coene
2001-06-15 5:57 ` Stef Coene
2001-06-15 14:27 ` Ramin Alidousti
2001-06-18 5:40 ` Stef Coene [this message]
2001-06-18 5:40 ` Stef Coene
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
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=marc-lartc-99442832205183@msgid-missing \
--to=stafke@iname.com \
--cc=lartc@vger.kernel.org \
/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