WAN Congestion and Queuing

[Andrew Grech]

Hi Experts,

If there is congestion from a bidirectional Iperf test using the default fair queue class and I have a loss intolerant application that is bursty (transmits once every 5 minutes, 60 bytes then backs off) set to use the priority queue. Is there a period of time where I may see traffic drops or delay, As the link is already congested?  In this situation the default queue is not shaped or policed, it simply has a minimum bandwidth allocation (which only really takes affect during congestion of higher classes)

One thing to note in this situation the headend router is 20mbps and tail only 512kbps. This is a bdsl circuit so the provider is performing qos outbound from their PE.

HQ Router QOS Markings> -------20Mbps--------ISP PE>QOS out>------512kbps-----<Branch Out QOS [Router]

A easy fix for this situation of the default queue congesting the link seemed to shape that queue so I had bandwidth reserved for the priority class. As the priority and bandwidth commands only take affect while their is congestion and that class is in use. However it felt like I was cheating as with rose colored glasses the priority queue is supposed help guarantee delivery.  

[Joseph W. Doherty]

Disclaimer

The Author of this posting offers the information contained within this posting without consideration and with the reader's understanding that there's no implied or expressed suitability or fitness for any purpose. Information provided is for informational purposes only and should not be construed as rendering professional advice of any kind. Usage of this posting's information is solely at reader's own risk.

Liability Disclaimer

In no event shall Author be liable for any damages wha2tsoever (including, without limitation, damages for loss of use, data or profit) arising out of the use or inability to use the posting's information even if Author has been advised of the possibility of such damage.

Posting

There's a couple of ways to deal with your topology, from HQ to branch.

You can shape for 512 Kbps, and under the shaper, prioritize traffic as desired.

You can shape for 20 Mbps (assuming the physical interface is "faster"), and under the shaper, prioritize traffic as desired.  Additionally, you need to allow for whatever QoS the provider has, and use that.

Of the two approaches, you would normally be better off shaping for your branch bandwidth.  Then you don't need to depend on your provider's QoS.

BTW, I believe many Cisco shapers don't account for L2 overhead.  If yours doesn't, then you need to shape slower to account for that overhead (often 10 to 15% slower works well).