Re: [Ietf-behave] multicast-04: need WG discussion
Hi,
(As Individual)
I think it would be a mistake to scrap this document. Multicast is
seeing more and more actual usage, especially in local access networks.
One such application is IPTV, where the IPTV top-set box preferably can
work across your home NAT. Also I don't share your thinking that RFC
4605 will be sufficient. I think it is important to include in the
BEHAVE series of document to also discuss the need to have a IGMP Proxy
in the NAT. In addition I think the single hard requirement included
within this specification is fully sufficient for publishing this document.
A few comments on the document.
1. Why did you remove the formulation that a RFC 4605 was mandated to be
supported.
2. In section 3:
" To prevent this unnecessary RTP collision detection by other session
members, the other session members need to see the same source
transport address for the RTP and RTCP traffic from the NAPTed host.
This requires the NAPT to assign the same UDP source port for that
RTCP traffic. This requirement also facilitates other, non-RTP
multicast applications which may function similarly."
I think the second sentence is wrong and confusing: "This requires the
NAPT to assign the same UDP source port for that RTCP traffic." Although
RTP and RTCP should have the same source IP address, they shall not
share UDP port. If they are sent on different ports from the internal
host, they should go out on separate ports.
Cheers
Magnus Westerlund
Multimedia Technologies, Ericsson Research EAB/TVA/A
----------------------------------------------------------------------
Ericsson AB | Phone +46 8 4048287
Torshamsgatan 23 | Fax +46 8 7575550
S-164 80 Stockholm, Sweden | mailto: magnus.westerlund@xxxxxxxxxxxx
