Linux CBQ question

From: Robert Cohen (robert@apex.net.au)
Date: Mon Jan 10 2000 - 03:31:45 EST


I'm trying to use CQB with Linux 2.2.14 to use both a tbf qdisc to do
packet shaping and a sfq qdisc to do maximise fairness.

The application is that there will be a 100 or so TCP connection active
and I want to limit aggregate bandwidth in as fair a manner as possible.

The script I am using follows.

tc qdisc add dev eth0 root handle 1: cbq bandwidth 10mbit avpkt 1000 mpu
64 allot 1514

tc class add dev eth0 parent 1:0 classid :1 est 1sec 8sec cbq bandwidth
10mbit rate 10mbit allot 1514 avpkt 1000 mpu 64 maxburst 50 split 1:0
prio 7 defmap 3f
tc qdisc add dev eth0 parent 1:1 sfq quantum 1514b perturb 15

tc class add dev eth0 parent 1:1 classid :2 est 1sec 8sec cbq bandwidth
10mbit rate 10mbit allot 1514 maxburst 50 avpkt 1000 mpu 64 split 1:1
prio 7 defmap 3f
tc qdisc add dev eth0 parent 1:2 tbf rate 245kbit buffer 2kb limit 50kb

I am attempting to load the sfq onto one class.
Then load the tbf onto another class with the sfq as its parent.

However this doesnt seem to work as the bandwidth doesnt end up limited.
If I load only the tbf it seems to work ok.

I have to admit that my understanding is a bit hazy as regards prio,
defmap and split.
For example it seems like you always use split on your parent, so why
does the option exist at all?

Also I understand almost none of the options that are being used. They
are there because I copied them from other examples.
Suggestions are welcome as to whether what I am doing is sensible at
all.
Also appropriate settings for sfq quantum and perturb.
Also maxburst settings.

Can anyone shed some light on this.

--
Robert Cohen - Network Administrator      
Apex Internet     
robert@apex.net.au     http://www.apex.net.au
Ph (02) 6247 2000      Fax: (02) 6247 2711

- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/



This archive was generated by hypermail 2b29 : Sat Jan 15 2000 - 21:00:16 EST