These are the results of the tests I did with CBQ classes, subclasses and filters.
Used script (download)
RATE_TOT=140kbps IP=kriek RATE1=100kbps PRIO1="prio 3" DEV="dev eth0" OPTION="allot 1514 maxburst 20 avpkt 1000" tc qdisc del $DEV root tc qdisc add $DEV root handle 10: cbq bandwidth 10mbit avpkt 1000 tc class add $DEV parent 10:0 classid 10:1 cbq bandwidth 10mbit rate $RATE_TOT $OPTION bounded prio 3 tc class add $DEV parent 10:1 classid 10:10 cbq bandwidth $RATE_TOT rate $RATE1 $WEIGHT1 $OPTION $PRIO1 bounded tc filter add $DEV parent 10: protocol ip prio 3 handle 1 fw classid 10:1 # red filter tc filter add $DEV parent 10:1 protocol ip prio 3 handle 1 fw classid 10:10 # green filter iptables -F iptables -X iptables -N acc_0 iptables -N acc_1 iptables -A OUTPUT -t mangle -p tcp --dport 2000 -j MARK --set-mark 1 iptables -A OUTPUT -p tcp --dport 2000 -j acc_0
This is the basic setup of my tests. The red, green and blue lines represent the different filters. I use fw filters based on marking packets with ipchains.
The first tests I did, was trying to understand the different filters. These are the results (100kbps gives a real bandwidth of 75kbps.
| filter | class 10:10 | |
|---|---|---|
| bounded | bounded + isolated | |
| red | 157 kbps | 157 kbps |
| red + green | 109 kbps | 109 kbps |
| blue | 109 kbps | 109 kbps |
| green | Full | Full |
For the next test, I removed the bounded parameter of class 10:10.
Used script (download)
RATE_TOT=140kbps # In KiloBytes/Sec IP=kriek RATE1=100kbps PRIO1="prio 3" RATE2=30kbps PRIO2="prio 3" DEV="dev eth0" OPTION="allot 1514 maxburst 20 avpkt 1000" tc qdisc del $DEV root tc qdisc add $DEV root handle 10: cbq bandwidth 10mbit avpkt 1000 tc class add $DEV parent 10:0 classid 10:1 cbq bandwidth 10mbit rate $RATE_TOT $OPTION isolated bounded prio 3 tc class add $DEV parent 10:1 classid 10:10 cbq bandwidth $RATE_TOT rate $RATE1 $WEIGHT1 $OPTION $PRIO1 isolated tc filter add $DEV parent 10: protocol ip prio 3 handle 1 fw classid 10:1 tc filter add $DEV parent 10:1 protocol ip prio 3 handle 1 fw classid 10:10 iptables -F iptables -X iptables -N acc_0 iptables -N acc_1 iptables -A OUTPUT -t mangle -p tcp --dport 2000 -j MARK --set-mark 1 iptables -A OUTPUT -p tcp --dport 2000 -j acc_0
| filter | class 10:10 | |
|---|---|---|
| isolated | ||
| red | 157 kbps | 157 kbps |
| red + green | 157 kbps | Full |
| blue | 157 kbps | Full |
| green | Full | Full |
When class 10:10 is not bounded, the results are like expected. But when you isolated class 10:10 and create some traffic in it with a filter, the bandwidth is totally used like there is no QOS.
I add a second class as a subclass of 10:1. I want to put the traffic in the different classes, but I want the most traffic in class 10:10. So I give class 10:10 a rate of 130kbps and class 10:20 10kbps.
Used script (download)
RATE_TOT=140kbps IP=kriek RATE1=130kbps PRIO1="prio 3" RATE2=10kbps PRIO2="prio 3" DEV="dev eth0" OPTION="allot 1514 maxburst 20 avpkt 1000" tc qdisc del $DEV root tc qdisc add $DEV root handle 10: cbq bandwidth 10mbit avpkt 1000 tc class add $DEV parent 10:0 classid 10:1 cbq bandwidth 10mbit rate $RATE_TOT $OPTION bounded prio 3 tc class add $DEV parent 10:1 classid 10:10 cbq bandwidth 10mbit rate $RATE1 $WEIGHT1 $OPTION $PRIO1 isolated tc class add $DEV parent 10:1 classid 10:20 cbq bandwidth 10mbit rate $RATE2 $WEIGHT2 $OPTION $PRIO2 tc filter add $DEV parent 10:0 protocol ip prio 3 handle 1 fw classid 10:1 tc filter add $DEV parent 10:0 protocol ip prio 3 handle 2 fw classid 10:1 tc filter add $DEV parent 10:1 protocol ip prio 3 handle 1 fw classid 10:10 tc filter add $DEV parent 10:1 protocol ip prio 3 handle 2 fw classid 10:20 iptables -F iptables -X iptables -N acc_0 iptables -N acc_1 iptables -A OUTPUT -t mangle -p tcp --dport 2000 -j MARK --set-mark 1 iptables -A OUTPUT -t mangle -p tcp --dport 2001 -j MARK --set-mark 2 iptables -A OUTPUT -p tcp --dport 2000 -j acc_0 iptables -A OUTPUT -p tcp --dport 2001 -j acc_1
| filter | class 10:10/class 10:20 | ||
|---|---|---|---|
| isolated | 1/2 isolated | ||
| red | 79 + 79 | 79 + 79 | 79 + 79 |
| red + green | 145 + 13 | Full | Full (X) |
| blue | 145 + 13 | Full | Full (X) |
| green | Full | Full | Full |
Full (X): Non-isolated alone: 157 kbps
Isolated class alone: Full
2 classes together: Full with class 10:10 72%
To have better results than with the tests 2 and 3, I attached a second qdisc to class 10:1. This has the advantage that I don't have to attach a filter to a class and don't have subclasses.
Used script (download)
RATE_TOT=140kbps IP=kriek RATE1=130kbps PRIO1="prio 3" RATE2=10kbps PRIO2="prio 3" DEV="dev eth0" OPTION="allot 1514 maxburst 20 avpkt 1000" tc qdisc del $DEV root tc qdisc add $DEV root handle 10: cbq bandwidth 10mbit avpkt 1000 tc class add $DEV parent 10:0 classid 10:1 cbq bandwidth 10mbit rate $RATE_TOT $OPTION prio 3 bounded tc qdisc add $DEV parent 10:1 handle 20: cbq bandwidth $RATE_TOT allot 1514 avpkt 1000 tc class add $DEV parent 20: classid 20:10 cbq bandwidth $RATE_TOT rate $RATE1 $WEIGHT1 $OPTION $PRIO1 bounded tc class add $DEV parent 20: classid 20:20 cbq bandwidth $RATE_TOT rate $RATE2 $WEIGHT2 $OPTION $PRIO2 isolated bounded tc filter add $DEV parent 10: protocol ip prio 3 handle 1 fw classid 10:1 tc filter add $DEV parent 10: protocol ip prio 3 handle 2 fw classid 10:1 tc filter add $DEV parent 20: protocol ip prio 3 handle 1 fw classid 20:10 tc filter add $DEV parent 20: protocol ip prio 3 handle 2 fw classid 20:20 iptables -F iptables -X iptables -N acc_0 iptables -N acc_1 iptables -A OUTPUT -t mangle -p tcp --dport 2000 -j MARK --set-mark 1 iptables -A OUTPUT -t mangle -p tcp --dport 2001 -j MARK --set-mark 2 iptables -A OUTPUT -p tcp --dport 2000 -j acc_0 iptables -A OUTPUT -p tcp --dport 2001 -j acc_1
| filter | class 10:10/class 10:20 | ||
|---|---|---|---|
| isolated | 1/2 isolated | ||
| red | 145 + 13 | 145 + 13 | 145 + 13 |
| blue | Full | Full | Full |
You get the same results if you use a filter that points directly to the good subclass (blue filter) as if you use different filters to get there (red filters).
When you use an isolated class, the traffic is not putted in the classes like you would expect. Just don't use the isolated parameter and everything will be fine.