>>	i just do not feel like adding more complexity into it.  do you have
>>	a plan for real use of "N events in M second"?  if that can be
>>	covered by "N events in 1 second", I'd say i like simpler one.
>Well, I guess I had been thinking about icmp.errratelimit.
>Right now it's set to a 1000us interval, so perhaps
>we would set it to 1000pps, and I guess that's fine.

	two comments:
	- icmp{,6}.errratelimit is broken anyways due to limited timer
	  resolution (hz = 100, so any value under 10ms does not do the right
	  thing).  we should set icmp{,6}.errratelimit to 0 by default
	  (and add pps-based one or some other ways as a replacement),
	  or we should remove icmp{,6}.errratelimit completely.
	- "1000us interval" and "1000pps" are not exactly the same.

itojun