<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=gb2312">
<META content="MSHTML 6.00.2800.1491" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#d4d0c8>
<DIV><FONT size=2>Hello all, I am testing CBQ and HTB these days and I got a
lots problem which needs your help. It is a little long text :) Thank
you! (All my test is based on Redhat Linux9.0)</FONT></DIV>
<DIV><BR>1.Is isolated of CBQ working? I have read Mr. stef word that he never
made cbq isolated work. </DIV>
<DIV><FONT size=2></FONT> </DIV>
<DIV>2. I have setup an environment to test cbq: one tcp flow which I should
protect<BR>and a noisy udp flow who will eat much of the bandwith.<BR><BR>when
only tcp flow is run, the result is good, the tcp rate is controlled with the
acceptable deviation.<BR><BR>But if the nosiy udp runs up(it is not classed into
the same class as tcp's)<BR>the tcp rate will increase almost to the double.Why
increased???Actually, I thought that tcp rate would decrease for impolite UDP
flow, but the fact is on the contrary. Whatever I set "isolated" or not to the
tcp rate.<BR><BR>The below is the script I set in a computer running redhat9,
and this linux also produces tcp flow to 200.200.200.229 and udp flow to
200.200.200.121 by iperf1.7.0. <BR>----------------------------------<BR>tc
qdisc add dev eth0 root handle 1:0 cbq bandwidth 100Mbit avpkt 1000 cell
8<BR><BR>tc class add dev eth0 parent 1:0 classid 1:1 cbq bandwidth 100Mbit rate
100Mbit weight 10Mbit prio 1 allot 1514 cell 8 maxburst 20 avpkt 1000
bounded<BR><BR>tc class add dev eth0 parent 1:1 classid 1:2 cbq bandwidth
100Mbit rate 480Kbit weight 48Kbit prio 1 allot 1514 cell 8 maxburst 20 avpkt
1000<BR><BR>tc class add dev eth0 parent 1:1 classid 1:3 cbq bandwidth 100Mbit
rate 10Mbit weight 1Mbit prio 1 allot 1514 cell 8 maxburst 20 avpkt 1000 bounded
isolated<BR><BR>tc qdisc add dev eth0 parent 1:3 handle 30: sfq<BR><BR>tc filter
add dev eth0 parent 1:0 protocol ip prio 1 u32 match ip dst 200.200.200.121
match ip dport 5001 0xffff flowid 1:2<BR><BR>tc filter add dev eth0 parent 1:0
protocol ip prio 1 u32 match ip dst 200.200.200.229 match ip dport 5001 0xffff
flowid 1:3<BR>--------------------------------------<BR><BR>3. I have watched
that cbq throughput is mostly 18M in redhat9. Although I wirte a script increase
"cbq bandwidth 100Mbit rate xxx" to be a larger and larger rate, but the iperf
reports only 18.5M or so. But HTB is very good at
it.<BR><BR>Thanks!<BR><BR>----- Original Message ----- <BR>From: "Stef Coene"
<<A href="mailto:stef.coene@docum.org">stef.coene@docum.org</A>><BR>To:
"cai" <<A href="mailto:rjcai@utt.com.cn">rjcai@utt.com.cn</A>><BR>Sent:
Wednesday, June 01, 2005 7:15 PM<BR>Subject: Re: cbq question<BR><BR><BR>> On
Wednesday 01 June 2005 12:51, you wrote:<BR>> > hello, stef, I saw your
assertion that cbq is not working in isolated mode:<BR>> > <A
href="http://mailman.ds9a.nl/pipermail/lartc/2002q3/004782.html">http://mailman.ds9a.nl/pipermail/lartc/2002q3/004782.html</A><BR>>
> "Btw, I never could get isolated working with cbq."<BR>> > Is this
true in all linux kernel version or only some specific.<BR>> It's a long time
since I tested it, but I never could get it working. The cbq <BR>> code
has not changed much, so I think it's stil not working.<BR>> <BR>>
<BR>> Stef<BR>></DIV></BODY></HTML>