Fco. J. Montilla wrote:

> Mensaje citado por Mayhem & Chaos Coordinator <[EMAIL PROTECTED]>:
>>The proper way to solve this is to use icecast as a tunneling mechanism.
>>If you have an Obs box at location A, and you want to listen to the same
>>stream at location B, but A and B are seperated by the Internet, then
>>set up icecast as a tunnel. You'll need an icecast server at both points
>>-- just have the icecast server at point B propagate a stream from
>>icecast server A.
>>
>>However, this isn't quite what you want. Icecast does not do multicast,
>>so location B cannot use multicast. Also, location B will not be able to
>>listen to the same stream as location A -- you will need to set up
>>different channels.
> 
> I went this route, and found that. No matter how I hack it, I wasn't able to
> stream/multicast the same channel. There's the other problem: I cannot once on B
> LAN multicast again the icecast stream.


I would have thought that rather than tunneling the data at the 
application level (which you say doesn't do what you want), a better 
solution is to tunnel the multicast traffic, so that you have two 
isolated multicast "clouds" connected over your VPN by TCP tunnels.  How 
to set this up depends a lot on your network setup, but you should be 
able to set up a BSD or Linux box as a multicast router using mrouted 
(you will need to enable multicast routing in the kernel, and set up all 
the tunnels in the mrouted config file).


> Another problem I'm facing is that all my mp3s are encoded at 128Kbps VBR with
> lame. The ADSL lines that interconnects some of the other branches are only 128
> upstream, so they can't cope with the variable 128 Kbps stream. 



mrouted can't help you here... there's already a bandwidth overhead 
stuffing the datastream into UDP multicast packets.  Stuffing those UDP 
packets into TCP packets to tunnel them only adds to the overhead...

Pete.

_______________________________________________
Obs-dev mailing list
[EMAIL PROTECTED]
http://www.freeamp.org/mailman/listinfo/obs-dev

Reply via email to