Re: [Cerowrt-devel] treating 2.4ghz as -legacy?

[Michael Richardson <mcr@sandelman.ca>]

Toke Høiland-Jørgensen <toke@toke.dk> wrote:
    >> I believe that Linux allows having both tagged and untagged packets on
    >> the samy physical interface, so the APs could communicate on a VLAN
    >> and one could be the gateway to the rest of the network (similar type
    >> of overhead in this case to GRE tunnels in that all traffic would get
    >> routed through one system, but I think it would still be less)

    > What happens to the VLAN tags if the traffic goes through a
    > non-VLAN-aware switch?

A switch which is totally ignorant of VLAN will simply pass all traffic to
all ports (modulo mac learning).  This is distinguished from a VLAN aware
switch that has it turned off (common for netgear equipment), which might
intolerant.

For many homes, the 3800 may be the only physical switch... a homenet router
discovery protocol would also notice if 3800s were directly attached...
The layer-2 VLAN idea fails if we have a routed homenet and there are a
variety of devices invoved.  That's why, I think that GRE over IP would win
for most homes, and yet keep the wired multicast and wireless multicast
separate.

--
]               Never tell me the odds!                 | ipv6 mesh networks [
]   Michael Richardson, Sandelman Software Works        | network architect  [
]     mcr@sandelman.ca  http://www.sandelman.ca/        |   ruby on rails    [