Larry,

Glad to learn that you agree that the NVE to NVA control plane requirement is 
exactly the same regardless of whether the NVE is embedded in Hypervisor, in 
End Device, or on a separate physical switch.

Then, I would suggestion re-wording the second paragraph of Section 3 to the 
following to avoid other people from getting confused:

There are two aspects of control plane for NVE: one is the protocol between NVE 
and NVA for NVE to be populated with the mapping tables  to tunnel traffic 
across the underlying networks; another one is the protocol between Tenant 
Systems and NVE for NVE to be promptly updated of TS's status. The latter 
control plane aspect is not discussed in this document, but is covered in 
[I-D.kreeger-nvo3-hypervisor-nve-cp-req]. The requirements for the NVE to NVA 
control plane are exactly the same regardless of whether the NVE is embedded 
within and End Device (e.g. Hypervisor), or in a physical switch.


I suggest to either remove the Figure 1, or change Figure 1 to show that NVE 
can be either embedded or separate:

[cid:[email protected]] 
--------[cid:[email protected]]

Figure 2 shows the Net Service Instances being attached to NVE. How is it 
different from VMs being attached to NVE from control plane perspective?

Linda


LK> The requirements for the NVE to NVA control plane are exactly the same 
regardless of whether the NVE is embedded within and End Device (e.g. 
Hypervisor), or in a physical switch.  The second paragraph of section 3 is 
trying to say that the Hypervisor to NVE protocol is covered in 
draft-kreeger-nvo3-hypervisor-nve-cp-req, not that that embedded NVEs is out of 
scope for the NVE to NVA control plane.  Please let me know how I can reword it 
to make it more clear.

<<inline: image005.png>>

<<inline: image006.png>>

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

Reply via email to