Teams QoS configuration

[ergamusai]

I need to configure QoS to prioritize Teams traffic. Is this config doing what I need to do

policy-map PM_MS_TEAMS_OUT
class CM_MS_TEAMS_AUDIO
priority level 1
police rate percent 10
class CM_MS_TEAMS_VIDEO
priority level 2
police rate percent 20
class CM_MS_TEAMS_SHARING
bandwidth remaining percent 15
queue-buffers ratio 10
class class-default
bandwidth remaining percent 35
queue-buffers ratio 25

policy-map PM_MS_TEAMS_QOS_SHAPER_OUT
class class-default
shape average percent 90
service-policy PM_MS_TEAMS_OUT

[Joseph W. Doherty]

Possibly, logically.  Insufficient information for optimal non-PQ bandwidth and/or buffer ration allocations.

Also shaper and hierarchal policy needed?  90% is a surprising number for a shaper.  At that level, why isn't 100% available.

[ergamusai]

I am not sure if the hierarchies policy is needed. I don't understand the last part very well, which is why I'm asking. My goal is prioritize teams traffic during congestions

[Joseph W. Doherty]

If you need to shape, your basic structure is fine.

If you don't need to shape, your child policy would be your only policy.

[ergamusai]

Thank you, seeing that we are maximizing our circuit, shaping should help. But from what I'm understanding correctly the shape average value should be 100. 

[Joseph W. Doherty]

Shaping is used to allow you to manage a downstream bottleneck that you otherwise cannot manage.

For example, if a WAN provider connects with you using fast Ethernet but tells you they won't allow more than 50 Mbps, shaping would allow you to manage that 50 Mbps.

Yours is a similar situation?

[ergamusai]

Ah, I see. I am not in the same situation. My provider facing interface is 10g and my bw is 10g. This has been very helpful. Appreciate your help and quick replies

[Joseph W. Doherty · Accepted solution]

In that case, you don't need the shaper.  As the interface queues, you can manage dequeuing prioritization.

Something you might check is if the interface has a large tx-ring.  Only when FIFO queue overflows into the policy queues do you get their dequeuing sequence.

I.e., you may want to reduce the size of the tx-ring.