I wanted to test how the latency of a packet would be affected by rate limiting the throughput of an interface. I have an interface named veth-red. It will received ICMP packet and routes it to veth-green. The veth-green then responds with the ICMP respone.

I was sending an ICMP packet of 1500 byte and wanted to set the rate on veth-red to be 10000 bits per second, so that I will get the response after 1.2 seconds. I used the following

sudo tc qdisc add dev veth-red root tbf rate 10000bit burst 12kbit latency 5000ms

Unexpectedly, I received the ping response in 0.057 ms. Also as soon as I change the burst to some value under 12kbit, my response is failing. How should I set the rate properly to see my expected behavior.

I have posted this question in the askubuntu forum but did not get any answer or suggestion.

  • Sorry Roaima. It was a typo – Ashish Kurian Jul 10 '17 at 8:57

The Linux traffic shapers (and Ethernet) work a packet at time and do not increase the time it takes to a single packet, rather they increase the time between packets.

If you set the maximum burst to smaller than a packet then it can't send the packet.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.