Help in understanding conference mode vs peer-to-peer mode

[Nagendra Kumar1648581702]

Hello Team AntMedia

,

   In the ant-media ecosystem, everything goes via the server (as expected). So on tech architecture where does peer to peer differentiate. 

If I understand correctly ant-media server has two modes - play and publish. In conference mode, these stream ids are getting created internally and shared with participants with the data channels. 

In peer, to peer, we have a similar play and publish - one user is publishing and another user is receiving the stream and playing but here the server is involved. Peer 1 is taking to the server and Peer 2 is also talking to the server.

So isn't thins technically the same it's just that stream ids in the conference is getting created internally.

I am not doubting on the architecture it's just that documentation is not clear about this. Please help me understand this. 

Thanks

Nagendra

[Tahir]

Hi Nagendra,

 

Peer to peer architecture normally uses mesh structure and conference is similar to SFU, however, what I understand is that you are asking about one-to-many architecture (which is not peer-to-peer) vs. conference architecture, this is correct right?

In that case they are very similar in terms of infrastructure, you are right about it. 

Otherwise, peer-to-peer uses a mesh structure which is pretty complicated if the number of peers go higher. With mesh structure it is probably not possible to have more than 10 people connected since every participant is sending data to each other. and receiving data from each other again which means 2N transfer for N other participant On the other hand, in SFU structure, every participant only sends data to server and receives other streams from the server which makes N+1 transfer for N other participant. 

I hope this will be helpfull

Tahir