From: none (alsontra@hotmail.com)
Date: Sat Oct 09 2004 - 13:14:07 GMT-3
All,
Q1. With a two rate policer you can control traffic based upon a CIR and a
PIR. Traffic conforming to the CIR can be marked or transmitted. Traffic
exceeding the CIR, but not in excess of the PIR can be marked or
transmitted, dropped. And finally traffic exceeding the PIR can be marked,
transmitted or dropped.
Cisco documentation specifies that the BC and BE values are related to the
CIR and PIR respectively. In addition, these values are described as
optional. Are the two rate policer BC and BE values only used to fine tune
the policers? Say, to avoid TCP saw tooth transmission patterns (packet
drop, global sync behavior)?
Q2. When marking with CAR is it still a good idea to use the standard
policing BC and BE calculations:
normal burst = configured rate * (1 byte)/(8 bits) * 1.5 seconds
extended burst = 2 * normal burst
I know that these calculations are good for standard CAR based policing
configurations. However, I'm so sure about the calculations when CAR is
used for marking.
The CCO states, "A normal burst size, which determines how large traffic
bursts can be before some traffic is considered to exceed the rate limit. An
excess burst size (Be). Traffic that falls between the normal burst size and
the Excess Burst size exceeds the rate limit with a probability that
increases as the burst size increases. CAR propagates bursts. It does no
smoothing or shaping of traffic." Meaning the BC and BE are only fine tuning
elements when used in marking configurations, but are not optional. ????
Reference:
http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122cgcr/fqos
_c/fqcprt1/qcfcar.htm
http://www.cisco.com/warp/public/473/153.html
TIA
Alsontra
This archive was generated by hypermail 2.1.4 : Sat Nov 06 2004 - 17:11:45 GMT-3