Skip to contentSkip to conversation

TBF tunning

LARTC Maria Joana Urbano 2 participants 2 messages Nov 7, 2003

lore
Share & tools
lore
#0 lore raw

Hi all,

I am trying to tune tbf qdisc. My upstream configuration is as follows:

tc qdisc add dev $DEV root handle 1: cbq avpkt 1000 bandwidth 100mbit tc class add dev $DEV parent 1:0 classid 1:1 cbq rate ${UP_RATE}kbit \
allot 1500 prio 5 bounded
tc qdisc add dev $DEV parent 1:1 handle 10: tbf rate ${UP_RATE}kbit \
burst $BURST limit $LIMIT mtu $MTU
tc filter add dev $DEV parent 1:0 protocol ip prio 18 u323 match ip \
dst 0.0.0.0/0 flowid 1:1

I tried several combinations of BURST (from 10KB until there is no pkt drops at tbf) and LIMIT values (corresponding to latencies of 40ms, 70ms and 140ms). MTU is set to 1500, the MTU of the link. I tested this configuration with several combinations of TCP flows and UDP flows (seperately). However, the best value I got for throughput was about 90% of UP_RATE.

My question is: is it possible to get better results with tbf, with throughput values closer to UP_RATE? And what should be changed in the script above to get it better?

Thanks,
Joana
_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/

#1 lore raw
Quoted 19 lines · Maria Joana Urbano

Hi all,

I am trying to tune tbf qdisc. My upstream configuration is as follows:

tc qdisc add dev $DEV root handle 1: cbq avpkt 1000 bandwidth 100mbit tc class add dev $DEV parent 1:0 classid 1:1 cbq rate ${UP_RATE}kbit \
allot 1500 prio 5 bounded
tc qdisc add dev $DEV parent 1:1 handle 10: tbf rate ${UP_RATE}kbit \
burst $BURST limit $LIMIT mtu $MTU
tc filter add dev $DEV parent 1:0 protocol ip prio 18 u323 match ip \
dst 0.0.0.0/0 flowid 1:1

I tried several combinations of BURST (from 10KB until there is no pkt drops at tbf) and LIMIT values (corresponding to latencies of 40ms, 70ms and 140ms). MTU is set to 1500, the MTU of the link. I tested this configuration with several combinations of TCP flows and UDP flows (seperately). However, the best value I got for throughput was about 90% of UP_RATE.

Quoted 3 lines

My question is: is it possible to get better results with tbf, with throughput values closer to UP_RATE? And what should be changed in the script above to get it better?

You don't need a tbf qdisc if you have a bounded class. The traffic is already bounded by the cbq class.

Stef

Signature · 8 lines
[email protected]
 "Using Linux as bandwidth manager"
     http://www.docum.org/
     #lartc @ irc.openprojects.net

_______________________________________________
LARTC mailing list / [email protected]
http://mailman.ds9a.nl/mailman/listinfo/lartc HOWTO: http://lartc.org/