C40/C60 Auto-Downspeed Due To Packet Loss?

[Michael Mertens]

  We're working on our QoS architecture for our up-coming roll-out of Cisco video end-points. Most of the conference rooms will be out-fitted with C40/C60s. We have an MPLS network and are trying to determine queue sizes for these components. Our hope is that if they experience packet loss (due to bandwidth or queue  constraints), they'll decrease resolution/down speed their bandwidth consumption till the packet loss comes into acceptable levels.

Our thought is to derive a policy of lowest accetpable level of resolution, and multiple that bandwidth by the max number of video calls at a given time, and that would equal our AF41 queue. (We'll have upgraded to CUCM 9.X and will have this value = video bandwidth LOCATIONS value).

Does anyone have any thoughts on this? Does this sound reasonable and doable with the c40/c60 behavior?

Thanks for any/all input.               

[Bhaskar Jha]

Yes, this should work fine. I will request to just refer the Admin and API guide for new ranges of ports and QOS vaules.

In TC6 there are few changes, refer to the guide of the S/W that endpoint will run on.

Thanks

Bhaskar

[Martin Koch]

The tagging part at least shall work fine, I am not so sure about the idea to accept loss and downspeed,

especially in both directions. You better have enough bandwidth or use a lower bitrate from the start.

The endpoints shall have a option:

xConfiguration Experimental Conference 1 ReceiverBasedDownspeeding: On

which is not very well documented and anyhow marked experimental.

So better not base your deployment on it ;-)

There are also:

*c xConfiguration Conference 1 PacketLossResilience Mode: On

*c xConfiguration Experimental Conference 1 PacketLossResilience ForwardErrorCorrection: OnWithUniqueSsrc

*c xConfiguration Experimental Conference 1 PacketLossResilience RateAdaption: On

but if I recall it right that will add additional traffic, so might cause more trouble when you do not have enough bandwidth