Darren.Reed at Sun.COM wrote:
> If you could connect your laptop to SWAN via both ethernet and
> WLAN using punchin, is there any reason why the command to
> create tunnel should be any different (aside from the IP addresses)?

No, there wouldn't be.

> 
> Ideally I'd be able to have the same IP address for both the LAN
> and WLAN connection (maybe it would be assigned to me by an
> 802.1x box) but that's probably wishful thinking.  But it would
> allow the tunnel to survive while I change from one network
> medium to another.

Okay.

> 
> The latter is something that's perhaps more easily done at
> home, if you have the same box issuing DHCP addresses for
> your WLAN as well as LAN, where you might desire to use the
> same IP address for both MAC addresses as they belong to
> the same laptop.
> 
> I'd almost go so far as to say that having any hard wired
> association between a tunnel and a link device, whether it
> is in the kernel or supplied by the user, is likely to get
> in the way of making things easy to use in the future.

But I don't see what this has to do with this discussion...

> 
> Which leads me to wonder if there will be a functionality
> regression here by requiring that there be a fixed association
> between a link device and a tunnel?

There is no such association on the table, and I wouldn't suggest that 
there be one.  The question is whether or not to allow something like:

dladm create-iptun -s <src> -d <dst>

and have dladm create a tunnel named "ip.tun0" or "ip.tun1" (or whatever 
unused name it chooses is appropriate), or whether the administrator 
should be required to type:

dladm create-iptun -s <src> -d <dst> <name>

... where <name> is "punchin0" or "mytunnel3", or whatever the 
administrator wants to name his tunnel.  Same question goes for create-vlan.

-Seb

Reply via email to