As stated in your Future Planing thread , you plan to integrate an MCU into your backend.
As a MCU will combine all streams and mix them together, it will also be possible to cancel noise (like echos, feedback). Will that be case?
An MCU will definitely need more server resources than a meshed network. Can you estimate how requirements will change for certain use cases?
Will it be possible to choose mesh over MCU for one on one calls only?
Is there a reason why you favour an MCU over an SFU?