[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
(usagi-users 02924) Re: IPv6 Multicast on recent Linux Kernels
On Wed, Apr 21, 2004 at 04:12:38PM +0900, YOSHIFUJI Hideaki / ?$B5HF#1QL@ wrote:
> In article <20040421065929.GA7866@xxxxxxxxxxxxxxxxxxxxxx> (at Wed, 21 Apr 2004 08:59:29 +0200), Stig Venaas <Stig.Venaas@xxxxxxxxxx> says:
>
> > On Wed, Apr 21, 2004 at 02:46:40PM +0900, YOSHIFUJI Hideaki / ?$B5HF#1QL@ wrote:
> > > In article <OF632B4C9E.CE204BC4-ON88256E7D.001B4C0D-88256E7D.001B7716@xxxxxxxxxx> (at Tue, 20 Apr 2004 22:59:58 -0600), David Stevens <dlstevens@xxxxxxxxxx> says:
> > >
> > > > A multicast router should not be needed if you're talking about
> > > > groups that should be forwarded by the HA, but I don't think
> > > > that'll work with or without a multicast router right now.
> > >
> > > Multicast router SHOULD be required to "forward" multicast packets.
> >
> > There are two implementations for IPv6, one written for exactly this
> > purpose, of so called MLD/IGMP proxies. In addition to just proxying
> > the MLD/IGMP, they also forward packets on the interfaces the joins
> > came from. They are both user space implementations. They could
>
> "Proxy"ing and "Forward"ing are different.
The term MLD proxy is a little bit strange and misleading. As I said MLD
is proxied, but part of the MLD (and IGMP) proxy concept is also to do
forwarding of multicast packets. Which interfaces to forward an incoming
packet on is, determined by which interfaces have received listener
reports. Many cheap home routers offer such proxying (and forwarding)
instead of using PIM or other multicast routing protocols.
See also draft-ietf-magma-igmp-proxy-06.txt on IGMP/MLD proxying. The
title is: IGMP/MLD-based Multicast Forwarding ("IGMP/MLD Proxying").
The multicast behaviour suggested by the HA, sounds pretty much the
same as this type of "proxying"...
Stig