Re: [Rmt] small group replication service, application-level multicast etc.

> 
>   But your system seems to be focused on the goal different from
>   ours. For real time datagram exchanging like VoIP and video
>   meeting, source of original is only one node and it is not
>   difficult, I think nearly impossible, for the listeners to
>   require other several nodes to relay each part of real time
>   streaming data depending of various unsettled bandwidth
>   situation.

While Swarmcast-like systems are optimized for static content 
distribution, they can certainly be applied to near-realtime delivery as 
well.

The tradeoff of using something like Swarmcast is that by encoding data, 
and delivering data out-of-order from multiple locations, you increase 
the latency of the communication.

So, for a near-live broadcast, you can easily buffer up 30 seconds of 
audio/video of data, and encode/interleave that chunk data swarming 
technologies.  So while the individual packets of the feed will be 
delivered out-of-order and from multiple sources in parallel, the 
overall feed can be reconstructed in its original sequential form.

So, while this approach obviously wouldn't work for a conference call 
VoIP application where latency is very critical, it might work well for 
a near-live news or sports broadcast where latency can be traded off for 
better performance and reliability.

>   For me, this looks the example of "There is no one-fits-all
>   multicast transport protocol."

So true.

-- 
Justin Chapweske, Onion Networks
http://onionnetworks.com/

_______________________________________________
Rmt mailing list
Rmt@ietf.org
https://www1.ietf.org/mailman/listinfo/rmt