[Serusers] Theory behind a media proxy and NATed UA

Nils Ohlmeier lists at ohlmeier.org
Thu Oct 6 13:51:42 CEST 2005


If one side of the connection is on the public internet and the other one 
behind NAT you do not need a RTP proxy in all cases.
But if both parties are behind NAT then you have a classical chicken-egg 
problem. For that scenarios you defiantly need a relay in the public 
internet.

  Nils

On Thursday 06 October 2005 13:17, Felipe Louback wrote:
> well, if after the NATed UA receives the 200 OK it has to start
> sending RTP and it is symetric(sends and receive data throught the
> same port), then there is no need of a media proxy, since the called
> part knows from which port in the NAT box the NATed UA is using...
> I am getting confused about why I need to use a media proxy....
>
> Felipe
>
> On 10/4/05, Felipe Louback <louback at gmail.com> wrote:
> > > after the UA, which send the INVITE, received the 200 OK for the INIVTE
> > > it has to start sending RTP.
> >
> > uhm..... I didn't know that. That is a very good explanation. If it
> > has to start sending RTP, then the media proxy now knows what port it
> > is using in tha NAT box. I thought that after the 200 OK, the nated UA
> > would wait for audio to come, since it sent all the parameters in the
> > SDP body.
> >
> > Thanks for the explanation.
> >
> > Felipe
> >
> > --
> > Master Student - Electrical Engineering Department
> > Computer Engineering and Telecommunications Research Group
> > Universidade Federal de Minas Gerais - Brazil
> >
> > "For God so loved the world that he gave his one and only Son, that
> > whoever believes in him shall not perish but have eternal life."
> > John 3:16
>
> --
> Master Student - Electrical Engineering Department
> Computer Engineering and Telecommunications Research Group
> Universidade Federal de Minas Gerais - Brazil
>
> "For God so loved the world that he gave his one and only Son, that
> whoever believes in him shall not perish but have eternal life."
> John 3:16




More information about the sr-users mailing list