Octave Orgeron writes:
> Hi James,
> 
> It's good to see this come up for integration.

Just to set expectations: we're actually nowhere _near_ integration.
That's months away.  We're still in an early stage of design and
development, and we're beginning to nail down some details.  If you're
interested in the project in more depth:

  http://www.opensolaris.org/os/project/rbridges/

> I was wondering how it'll impact virtualization (containers, xen, and ldoms)? 
> Will further work be required or will it work transparently with things like 
> VSW's in LDoms?

In general, it follows the same model as 802.3 aggregation.  Any place
that can be configured, bridging can be configured, and where it
can't, you won't have bridges.

For Zones, bridges won't be possible.  The default shared-stack type
of zone does not have access to datalink layer interfaces at all, so
bridging wouldn't apply.  For the exclusive-stack type of zone, the
current assignment logic places only VLAN objects inside the zone, and
not the basic links, so you'll have no way to construct a bridge.

(If you could do that, the semantics would be strange.  You'd end up
with one zone connecting interfaces together behind the collective
backs of the other zones, and little in the way of observability.  I
suppose it's possible, but the system [particularly the Nemo link
layer] isn't currently designed to support it.)

For Xen and LDOMs, I'd expect bridging to work normally, in as much as
those solutions give the guest operating system access to a MAC-layer
device of some sort.  Obviously, there are some permutations in here
that we'll need to include in our testing, but I don't see any
architectural issues that would preclude it.

-- 
James Carlson, Solaris Networking              <james.d.carlson at sun.com>
Sun Microsystems / 35 Network Drive        71.232W   Vox +1 781 442 2084
MS UBUR02-212 / Burlington MA 01803-2757   42.496N   Fax +1 781 442 1677

Reply via email to