Re: [OT] question on NANOG meetings
the it behaves well in varying network conditions, which cannot be said from either WM or RM.
Quality of the link from encoder to Real server is important, send garbage to the server and it has nothing to distribute. I get the impression it's done in band with other NANOG traffic and may be suffering against the warez traffic. Some details of that link and the server loading would be interesting, anyone got details from previous meetings? brandon
On Sat, 27 Sep 2003 14:46:37 +0100 (BST) Brandon Butterworth <brandon@rd.bbc.co.uk> wrote:
the it behaves well in varying network conditions, which cannot be said from either WM or RM.
Quality of the link from encoder to Real server is important, send garbage to the server and it has nothing to distribute.
I get the impression it's done in band with other NANOG traffic and may be suffering against the warez traffic.
Some details of that link and the server loading would be interesting, anyone got details from previous meetings?
FWIW, I try to look at the _multicast_ feedback as I can snarf it down here. Typically I get Multicast RTCP receiver reports from 10's of receivers, which suggests (as most players do not send RR's) a total multicast audience of order 100. At least some receivers tend to report low packet loss (typically order 1/2 of all reports say zero loss). Regards Marshall Eubanks
brandon
Brandon Butterworth wrote:
the it behaves well in varying network conditions, which cannot be said from either WM or RM.
Quality of the link from encoder to Real server is important, send garbage to the server and it has nothing to distribute.
One can screw it up earlier than that, send garbage to the encoder and it spends 90% of the available bandwidth encoding noise.
I get the impression it's done in band with other NANOG traffic and may be suffering against the warez traffic.
Diffserv anyone? Pete
participants (3)
-
Brandon Butterworth -
Marshall Eubanks -
Petri Helenius