Steffen Weiberle writes:
> > The part that's missing is allowing DHCP to be configured in zonecfg
> The zonecfg part works.


> > and correctly handled by zoneadmd.  I don't see how zoneadm itself
> > could be at all involved.
> zoneadm's verification fails.

Ah, ok -- it's probably zonecfg_valid_net_address.

> who does the work to 'ifconfig 
> addif' on behalf of a non-global zone when the IP address is set? 

zoneadmd does this.

> Is that not zoneadm*? It can't be the zone itself, since it does 
> not have the privileges.

Right.  The zoneadm utility uses a door call to ask the daemon to do
the startup/shutdown work.

> > I see.  I think that just pushes the issue elsewhere, as you have to
> > deal with (potentially) very large numbers of MAC addresses.
> Is the concern about DHCP handling that?


> With hardware rings and
> soft MAC, the number of MAC addresses is going to get large. And
> it may be necessary to be able to fully utilize 10GbE interfaces and
> still offer virtualization, flow management, etc.

The concern is that for VNICs (as best I understand), you need a
separate MAC address for each instance.  That's a potential
administrative burden if you have hundreds or thousands of these

The new DHCP logical interface support doesn't suffer from that
problem.  It's able to create unique identifiers without
administrative intervention.

So, yes, I think you could potentially solve that same problem of
having multiple zones on one physical interface using VNICs, and there
are probably some advantages to doing so, but one detraction would be
having to manage the MAC addresses.  Doing the same thing by using
DHCP's support for logical interfaces would avoid that particular

James Carlson, Solaris Networking              <[EMAIL PROTECTED]>
Sun Microsystems / 1 Network Drive         71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677
zones-discuss mailing list

Reply via email to