seeking clarification on what the audio/video streams would look like if you were to wireshark the traffic for a multi point video conference call.
example is a handful of remote sites with quickset C20's deployed
central site has a few endpoints, + VCS, TMS, and a 4501 bridge
question is this. from a network perspective - say 5 remote sites dial into a bridge
they are transmitting their audio/video across their wan link to the central site - this makes sense...all good.
say for example I have configured this bridge to be capped at 1150Kbps per participant.
at the central site - the MCU is receiving 5 streams of 1150Kbps from each participant. True?
what do the transmit streams look like from the MCU out to the remotes? if someone wireshared all this traffic - I am under the impression that you'd see a source/dest pair from the MCU to each particpant... so the outgoing bandwidth would be 1150Kbps * 5 particpants.
is this the right way to look at this?
The MCU looks like the center of all communication for this conference call - with video traffic being transmitted to each participant that appears as the "multiplexed" image of each particpant - with the speaking person being the center point.
The issue is Video unlike Voice isn’t deterministic. It depends on other factors such as movement, lighting etc. All of that have an effect in the BW, so if there are more or less participants in the room and depending on movement the BW will fluctuate. You can see in real time on all the end points what you are sending and receiving and you will even on a call with 384Kbps, the BW might come down to 150Kbps, then some burst as people enter the room, switch on the lights etc.
In addition as codec have gotten more sophisticated they only send frames as they change and then every now a complete refresh is done. So you should engineer for the worst case but in reality you will not consume 1.5 Meg all the time
As I have been told by Cisco engineers, if the MCU has 5 calls at 720p30 then you will need 5 x 1152kbps for downstream and 5 x 1152kbps for upstream.
Though, as Srini has said, BW will fluctuate depanding on number of factors.
Thanks for the feedback above.
this is how i understood it as well... I agree with first reply - i didnt mean to suggest that it would always be 1152K...but was seeking clarification - the bandwidth at the MCU is cumulative for each additional participant that joins the bridge call.
That is correct as a worst case number. In reality it will be much different but it’s always best to engineer for the worst case scenario