I see that in bandwidth plugin, we set an ifb and limit bandwidth of this ifb, but I have done a test that I enter into pod's network namespace and set tbf qdisc to the 'eth0' interface. In this way, we can limit pod's egress traffic and we don't need to set up an ifb in host network namespace. I am not sure if there is any concern that we use ifb in host network namespace instead of set egress limit to pod's eth0 in pod's network namespace ?
I see that in bandwidth plugin, we set an ifb and limit bandwidth of this ifb, but I have done a test that I enter into pod's network namespace and set tbf qdisc to the 'eth0' interface. In this way, we can limit pod's egress traffic and we don't need to set up an ifb in host network namespace. I am not sure if there is any concern that we use ifb in host network namespace instead of set egress limit to pod's eth0 in pod's network namespace ?