Traffic shaping and queueing in 7200 series VXR router

[vish_arun]

I have been doing some work using Cisco 7200 series VXR router. I have two questions on traffic shaping. Replies are much appreciated.

1. I am using class based traffic shaping using Cisco IOS 12.2. I create a class-map and then a policy-map to shape the average output interface rate

to 20Mbps (I am using the "shape average" command to do this). I used "set max-buffers 30" to set the queue length to 30.

I want to implement FIFO queueing. However, when I use the "show traffic-shape queue" command to view the shaping information, it tells me that the queueing strategy is Weighted Fair Queueing (WFQ). Can you kindly let me know how I can implement FIFO instead of

WFQ?

2. In the Cisco Internetworking Handbook, it is mentioned that a queue is created for EACH FLOW when

WFQ is enabled. It is also mentioned that "set max-buffers" specifies the maximum number of buffers allowed on shaping queues. Can you kindly help me

interpret the value of 30 above? Does it mean that there are 30 packets worth of buffering FOR EACH FLOW or something else? In other words, if there is 1 TCP

connection passing through the router, then there is 1 QUEUE with 30 packets of buffering; if there are 2 TCP connections, there are 2 QUEUES with 30 packets each and so on.

Many thanks !

-Arun

[sbilgi]

30 is the maximum buffer size.

Try this link

http://cisco.com/en/US/products/sw/iosswrel/ps1835/products_command_reference_chapter09186a0080087f36.html

[ronbuchalski]

In order to use FIFO queueing with a policy-map, you need to create a priority policy within the CBWFQ policy-map:

http://www.cisco.com/en/US/docs/ios/12_3/qos/command/reference/qos_o1g.html#wp1076758

The LLQ or Priority Queue will be a FIFO queue.

[Joseph W. Doherty]

#1

This might be accomplished with a child policy.

e.g.

policy-map aChild

class class-default

bandwidth ###

policy-map aParent

class class-default

shape average ###

service-policy aChild

#2

Documentation isn't clear for a shaper, i.e. whether setting is for total buffer space for per flow queue limit.