The L2 p2p link ends at the BTS (?) , but the L3 subnet ends at GGSN (?) which is a router. BTS supports multiple p2p links, any of them can have a mcast listener. So the BTS needs a way to figure out to which L2 p2p link a mcast packet needs to be forwarded. Without MLD support, how does the L2 switch know there is a mcast listener in the link so it can copy a packet to that link ? The only other way would be to manually configure the L2 switch - not scalable.
Subrata -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED]]On Behalf Of [EMAIL PROTECTED] Sent: Friday, May 24, 2002 8:36 AM To: [EMAIL PROTECTED]; [EMAIL PROTECTED] Subject: RE: MLD comments on cellular host drafts - seeking feedback Hesham, this was a good summary. So how about going for alternative 3? Cheers, -Juha W.- -----Original Message----- From: ext Hesham Soliman (ERA) [mailto:[EMAIL PROTECTED]] Sent: 23 May, 2002 18:34 I've tried to summarise the current situation with MLD and see how we can move forward with the next revisions of the draft. Jari will send another emails for the rest of the issues discussed. First let me summarise the comments: - MLD must be used for all multicast addresses except the all nodes multicast address. This is what RFC2710 says. My understanding (please correct me if I'm wrong) was that the need for using MLD for link-scope addresses was to allow for the cases where L2 switches snoop MLD reports. As far as the cellular host is concerned, we are dealing with p2p links where the neighbour is a default router. So for link local traffic in general the notion of multicast does not exist (even if, for example an RS is sent to the all routers multicast address, it is going to one router). And there are no switches that will need to snoop MLD traffic. Sending MLD reports for link local addresses, will not only increase the use of BW on the links but is unnecessary for this p2p link. However, MLD will be necessary for multicast addresses with scopes larger than the link scope. So based on the dicussion above, we have a number of choices for the next update: 1. Suggest that MLD must be used even for link local addresses in p2p links within cellular networks (3GPP). pros: We follow RFC 2710. cons: Waste of expensive BW unnecessarily 2. We suggest that for these p2p links, with no switches snooping MLD traffic, MLD should only be used for multicast addresses with scopes larger than link-local. pros: We save expensive BW cons: We don't follow RFC2710, but then again RFC2710 does not make a clear distinction between these p2p links and other multicast links. 3. We suggest that this exception (of not using MLD for link-local multicast traffic) can be made, until RFC2710 is updated to explain why the mandate was placed and whether there should be any exceptions to that mandate. pros: We also minimise the scope of the exception to 3GPP only. cons: We still don't follow this particular part of RFC2710 I look forward to your choices from the above selection :) or even new suggestions. Hesham -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPng Working Group Mailing List IPng Home Page: http://playground.sun.com/ipng FTP archive: ftp://playground.sun.com/pub/ipng Direct all administrative requests to [EMAIL PROTECTED] --------------------------------------------------------------------
