Cisco Support Community
Showing results for 
Search instead for 
Did you mean: 
New Member

LLQ dilemma about PQ

I am intensively preparing for the QoS rollout in my company (I'm a network admin) and I was reading many papers and books about it. Still, there's one thing left to clarify.

Is PQ policed or not? Some books and courses say "yes, definitely, even when there's bandwidth available" and some say that PQ will get the bandwidth above the priority value entered - if it's not taken by the other clasess.

For instance, the book "End to end QoS design" (really great reading, indeed) says that newer IOSes have this feature implemented, so the PQ can take more bandwidth if it's available.

What is the final truth? If the book mentioned is right, can someone tell me from which IOS version that feature is supported?


Re: LLQ dilemma about PQ


My experience says the CBWFQ priority queue is always policed regardless of available bandwidth. Just to be sure I set up a test in my lab, and was surprised by the results. At least in my limited test 12.4 code did allow the priority queue to use available bandwidth while 12.3T did not.

My setup is a 7206 running 12.3(14)T6 connected via T1 to a 2801 running 12.4(10a). I set the priority queue in the outbound service policies to 16kb, placed a G711 call, and generated a steady 256kb of default traffic in both directions. Leaving the 7206 the 30 second offered rate for the voice queue was 80000 bps, and the drop rate was 6000 bps.

The 2801 showed 80000 offered and no dropped packets. Total traffic rate was 336000 in each direction.

With default traffic of 2-3kb neither router dropped any voice packets.

While far from conclusive, my testing at least suggests that 12.4 code would be more forgiving when a link is not congested. Of course, CAC should always be configured to prevent your priority queue bandwidth from being exceeded.

Please rate helpful posts.


Hall of Fame Super Gold

Re: LLQ dilemma about PQ

Good info Dave, thanks.

CreatePlease to create content