|
Posted by glen herrmannsfeldt on May 1, 2008, 3:14 pm
Albert Manfredi wrote:
(snip)
>>Somewhere back in the origins of ethernet is the ability to
>>assign MAC addresses either to ports or hosts.
>>The only one I know to commonly apply them to hosts is Sun,
>>which used to put the address in ROM on the CPU board which
>>may or may not have an ethernet interface. That address was
>>then used on all ports on that host. (Sun machines were
>>often configured as routers with more than one ethernet port.)
> That was also my reaction to the original post. It is possible that
> the router might have only one MAC address assigned to it, if it
> follows that somewhat iffy (IMO) Sun philosophy you mention.
> Note this, however. Access Points are not necessarily combined with a
> router function. Therefore, the two interfaces of an access point
> *could* exist within the same network (IP subnet). Whereas the two
> interfaces of the router would, by definition, each be in its own IP
> subnet.
> Therefore, it's acceptable to assign just one MAC address to a router,
> but not to the access point.
> Another minor point is that sometimes switches use the MAC address of
> one of their interfaces as the MAC of the entire box, if a single MAC
> address to identify the box is needed. I'm too lazy to look it up
> right now, but doesn't RSTP do this, for example?
So it would be just a bridge. I have thought about connecting just
to the LAN port on a router/access point to put the WLAN on the same
subnet as an existing wired LAN.
Does a switch without spanning-tree need any MAC address? If so,
does it need more than one?
-- glen
But without a router function, why two addresses?
|