Sebastien Roy writes: > > On Thu, 2009-03-26 at 21:25 -0700, Peter Memishian wrote: > > > That indeed doesn't imply that one should be able to create a VNIC from > > > within a non-global zone, but it does hint at some requirements for link > > > namespaces in non-global zones as well as the idea that one will need > > > the ability to control the properties of a link in a non-global zone. > > > For example, we'll eventually need the ability to consolidate two > > > physical systems, each with two physical links named "net0" and "net1" > > > into two non-global zones on a single system. > > > > Except it will still be restricted -- e.g., you can do things like create > > aggregations and VLANs over the physical Ethernet links, but not over the > > VNICs that are assigned to the two non-global zones. In general, VNICs > > are a nice way to get a datalink into a zone for direct use by IP, but not > > a full substitute for a physical link. > > That's true. Along this line of discussion, would it not be a > reasonable feature to create VNICs over VNICs, and thus VLANs over > VNICs?
I can see how someone might want to do that, but I think that direction may have problems. VLANs, aggregations, port security, LLDP, and (for that matter) bridging are all 802 concepts. They have very clear and non-trivial relationships among one another. For one thing, they have an ordering (e.g., port security must be below aggregation, which must be below bridging, which must be below VLANs). For another, they have knowledge of one another (LLDP and bridging know about VLANs). Allowing arbitrary nesting will potentially break those relationships by permitting the user to set up configurations that not just violate the standards but also possibly result in network failure. For example, creating a VNIC on a bridged NIC and then a VLAN on top of that VNIC potentially leaves the bridge unaware that the VLAN exists, resulting in a loss of connectivity. Entities outside of our box (network peers) are unaware that we have these substructures inside, and I think we have to keep it that way. If we allow something other than the VNIC/VLAN/aggr/NIC relationship we have today, I think we'll need to enumerate the possibilities and work through all of the implications carefully. -- James Carlson, Solaris Networking <[email protected]> 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 _______________________________________________ networking-discuss mailing list [email protected]
