Sorry for jumping in late to this discussion.

I agree with Tom's definition.

In addition, I would like to point out a use case which is a hybrid.
We are doing a overlay deployment model for VM's in the data center at
PayPal. 
One of the issues that come up is to have a separation of production
traffic and management traffic at the VM level.
We use the management traffic for pushing patches, release updates and
Openstack control traffic. It is quite possible that the management NIC's
on the VM's needs to be bridged to a physical NIC on the hypervisor which
is used for management purposes of the hypervisor. In this case, the VM's
had two NIC's one of them using overlays on the production NIC and the
other NIC is bridged to the management physical NIC on the hypervisor.

-- 
Vinay Bannai
408-967-7784
Cloud Engineering







On 4/19/13 5:05 PM, "Thomas Narten" <[email protected]> wrote:

>> I don't agree that it always appears as a NIC.  It appears as some
>> sort of protocol interface, but that is not necessarily a NIC to the
>> tenant system.   For example, in the physical world, a VLAN doesn't
>> look like a NIC to the host OS.
>
>In Linux at least a VLAN does look like a NIC (interface) on the
>TS. You first configure the interface, then to use a VLAN, you end up
>creating a virtual interface (for that VLAN) that connects through the
>other interface. To the linux host, each VLAN appears as its own
>interface.
>
>See the "vconfig" command for instance on how this works.
>
>But in any case, I think the broader point stands that a TSI is the
>hypervisor-facing side, whereas on the TS, you get an interface of
>some sort. The exact details of how that interface is presented depend
>on the TS (whether linux, microsoft, etc.) itself. But that is the
>case whether the interface is physical or virtual.
>
>Thomas
>
>_______________________________________________
>nvo3 mailing list
>[email protected]
>https://www.ietf.org/mailman/listinfo/nvo3

_______________________________________________
nvo3 mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/nvo3

Reply via email to