From: Stef Coene <stef.coene@docum.org>
To: lartc@vger.kernel.org
Subject: Re: [LARTC] Fair Queuing
Date: Fri, 22 Nov 2002 17:06:11 +0000 [thread overview]
Message-ID: <marc-lartc-103798493517517@msgid-missing> (raw)
In-Reply-To: <marc-lartc-103797298703293@msgid-missing>
On Saturday 23 November 2002 06:58, ajay@movingdelhi.org wrote:
> Thanx, I managed it to get the basic script working but the result IS NOT
> satisfactory. What is happening is this:-
>
> When any user (ex 192.168.0.4 in my script) starts to download a file
> (Download1) after a while it settles down to the alloted bandwidth 32kbps.
> If he puts another download (Download 2) after a while it occupies
> approximately half of the total bandwidht alloted to 192.168.0.4. So far
> so good.....
> NOW, when the user(192.168.0.4) cancels any one of the downloads (Say
> download 1) the other download (download 2) takes a very long time to grow
> up to the total alloted bandwidth i.e 32kbps.
>
> Why is that happening?
I'm not sure, but when the scond download is started, both ends of the
connection are using as much bandwidth as possible untill packets are
dropped. If the connection has suddenly more bandwidth because the other
download is stopped, it takes some time before both ends of the connection
realize this. And when they do so, they are trying to push again as much as
possible untill they are throttled to 64kbps. I think that's the way how tcp
is handle the bandwidth limitaion.
> I can't figure it out. Could SFQ Help?
Not really. SFQ give each data stream the same opportunity to send something.
If one of the streams is gone, the other can send all of the time. But you
can try to add one to see what happens.
> ___________________________________________________________
> /sbin/tc qdisc add root dev eth1 handle 1:0 cbq bandwidth 10Mbit avpkt 1000
> /sbin/tc class add dev eth1 parent 1:0 classid 1:1 cbq bandwidth 10Mbit
> rate 256Kbit allot 1514 bounded
> /sbin/tc class add dev eth1 parent 1:1 classid 1:2 cbq bandwidth 256Kbit
> rate 32Kbit allot 1514 bounded
> /sbin/tc filter add dev eth1 parent 1:0 protocol ip prio 1 u32 match ip src
> 192.168.0.4 flowid 1:2
> /sbin/tc filter add dev eth1 parent 1:0 protocol ip prio 1 u32 match ip dst
> 192.168.0.4 flowid 1:2
> ________________________________________________________________
bandwidth is always the nic bandwidth so 10Mbit.
And shaping on eth1 is shaping all the packets that are leaving eth2. So the
usung src and dst in 2 filters is useless. The src filter will never match.
Stef
--
stef.coene@docum.org
"Using Linux as bandwidth manager"
http://www.docum.org/
#lartc @ irc.oftc.net
_______________________________________________
LARTC mailing list / LARTC@mailman.ds9a.nl
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/
next prev parent reply other threads:[~2002-11-22 17:06 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2002-11-22 13:49 [LARTC] Fair Queuing ajay
2002-11-22 15:07 ` Stef Coene
2002-11-22 16:35 ` ajay
2002-11-22 17:06 ` Stef Coene [this message]
-- strict thread matches above, loose matches on Subject: below --
2004-03-30 7:44 [LARTC] "Fair" queuing Mihai Vlad
2004-03-30 12:00 ` Jeroen Vriesman
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-103798493517517@msgid-missing \
--to=stef.coene@docum.org \
--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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.