[Live-devel] Streaming from client to server

[Jeff Shanab]

Thanks for this thread, I am also dealing with rtsp push but quite a different animal.

Traditional pull requires ports open thru firewalls and static IP's (or dyndns)
A push (or confiscated connection from initial push) is one way to stream from behind a firewall without setting up port forwarding.

From: live-devel-bounces at ns.live555.com [mailto:live-devel-bounces at ns.live555.com] On Behalf Of Ross Finlayson
Sent: Thursday, August 25, 2011 11:08 PM
To: LIVE555 Streaming Media - development & use
Subject: Re: [Live-devel] Streaming from client to server

Just because I'm curious (and I recognize this goes beyond support of the Live555 library so no problem if you don't have time to respond), could you expand on why you think the push model is a bad idea

The biggest problem that I see with the 'push' model is that it complicates the server implementation, because the arrival of the data into the server is decoupled from the streaming of the data out of the server (i.e., to regular clients).  I.e., an implementation of a server that supports the 'push' model will need to include some sort of buffering mechanism (to buffer the incoming data).  Then, when streaming this data out to clients, the server will read - i.e., 'pull' - from this buffer.  But because the server is, in reality, 'pulling' the data that it streams out to clients, then why not just have it 'pull' the data directly from the source?  This is much easier to implement.  (In fact, if the underlying OS supports a 'networked file' abstraction, then it's trivial to implement; the server can just continue to think that it's streaming from a file.)

Ross Finlayson
Live Networks, Inc.
http://www.live555.com/

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.live555.com/pipermail/live-devel/attachments/20110830/9483ded3/attachment.html>