Hi, Pat: I agree we should not restrict the number of physical NICs. I will fix this in the draft. I am just thinking whether we have a case where one tenant system both havs pNIC and havs vNIC facing the same side ,i.e., hypervisor side. Besides we have: a) a tenant system only having pNICs facing hypervisor side. b) a tenant system only having vNICs facing hypervisor side. It looks Larry give one example of such case, Referring to what Larry said in http://www.ietf.org/mail-archive/web/nvo3/current/msg02251.html " For example, if the TS is a router, it may use the burned-in MAC and also have a virtual MAC for supporting VRRP. If the TS is a bridge, it will send/receive frames with the MAC address of all the entities it is bridging traffic for. " Regards! -Qin -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Pat Thaler Sent: Saturday, April 20, 2013 2:30 AM To: Thomas Narten; Qin Wu Cc: [email protected] Subject: Re: [nvo3] vNICs and pNics in draft-wu-nvo3-nve2nve-04.txt
In addition to Thomas's point, we should not restrict the number of physical NICs that a tenant system can have. Some tenant systems will have more than one physical NIC. We may describe some typical tenant systems as part of examining use cases, but NVO3 should define behavior in terms of the network interface, i.e. TSI, behavior and should not restrict tenant system architecture. Pat -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Thomas Narten Sent: Friday, April 19, 2013 7:07 AM To: Qin Wu Cc: [email protected] Subject: [nvo3] vNICs and pNics in draft-wu-nvo3-nve2nve-04.txt Qin Wu <[email protected]> writes: > [Qin]: I agree with one tenant system may have one pNIC and one ore > multiple vNICs, Let me zero on in this because I don't quite understand this model, and I suspect this point is leading to the back-and-forth on the terminology thread. The two terms are also used in draft-wu-nvo3-nve2nve-04.tx, which says: > o Each tenant system is corresponding to one virtual machine. > Each tenant system has only one pNIC and one or more vNIC > adapters that it uses to communicate with both the virtual and > physical networks.The pNIC and vNIC adapters each virtual > machine has belong to a single tenant. To me, a Tenant System (TS) doesn't have pNICS and vNICs. It has NICs. By definition, a TS is connected to one or more virtual networks (VNs). If it has a native connection to the DC network, that is out-of-scope for NVO3. Plus, I don't know why it would do that, or what implication it would have for NVO3. To the TS, it has NICs, but it really doesn't know whether they are physical or virtual. The whole point is that the TS just uses the NICs it has as if they were physical. Hence, it doesn't make sense to talk about a TS having both kinds of connections. Shouldn't that just be completely transparent to the TS? Why are you distinguishing vNIC and pNIC in the context of a TS? 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 _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
