Hello Qiang, Thanks. I like to make one thing clear about this document. It is to specify the data plane NVE functionality which is necessary for TS-NVE and NVE-NVE interoperability. It leaves the NVE or NV03 control plane functionality out of the scope. The NVO3 framework and architecture documents, IMO, aim on the framework/architecture so we can describe the requirements under the reference model. They are not meant to specify the component functionality for the interoperability purpose. This document aims on this purpose. People may see some similar descriptions between these documents here and there, but this document has the functionality specification for the interoperability.
Please see inline. -----Original Message----- From: Zu Qiang [mailto:[email protected]] Sent: Wednesday, November 13, 2013 8:27 AM To: Lucy yong; [email protected] Subject: RE: New Version Notification for draft-yong-nvo3-nve-01.txt Hello, Lucy Nice document. It has many interesting text. A few clarification questions below. - Very first comment: it would be good to make it clear in the draft what is the working assumption of the NVE function. I see you have many "out of scope" in the text. I understand that how the information is configured in the NVE is not in the scope. But it would be good to make it clear the what. For instance, it is better to make it clear that "assuming the TS has a MAC address configured if L2 service is provided by the NVO3. The TS MAC address may be configured via the VM Orchestration system which is out of the scope of this document." [Lucy] NVE configuration can be done by manually or automatically. That all we need to say here. How it is configured should not impact the NVE functionality. The document should not mandate how TS configuration that you give. - 3.1: We cannot assume that the NVE always knows that the TS has at least one L2/L3 address configured, right? It shall be that TS may have a L2 address configured if L2 service is supported by the NVO3 and TS may have a L2 and L3 address configured if L3 service is supported by the NVO3 [Lucy] TS is ether configured with IP gateway address or not, in both case, TS use protocol like ARP to resolve Gateway MAC address. When L2/L3 NVE (i.e. distributed gateway function) is used, NVE has the MAC address for the distributed gateway and always responses that this MAC address in ARP reply regardless TS ARP request has IP gateway address or default address. NVE gets the TS gateway information from NVA. So it will forward the packets to the gateway when the packet has the destination IP addr as the gateway address. In this case, NVE serves as the first hop routing. If inter-VN is allowed at NVE, the NVE serves as the TS gateway role, directly forward the packet to another a VN on the NVE, the NVE performs the packet forwarding as of the packet from the second VN. - ARP is not the only way for peer MAC learning. Better to not limit ourselves to a pure IPv4 network . Needs to make it more generic to cover different network use cases, such as IPv6, 802.1, etc. [Lucy] agreed. We will describe other ways in next version. - same comments for the GW MAC address learning [Lucy] several common methods to achieve this. We do not need to invent another one. We just need to address how NVE to interwork with these methods in the draft. - Destination address caching, Ingress filtering and Forwarding handling in the TS, is this an implementation? Or is this just a working assumption for some error case handling? Please clarify it. [Lucy] I don't know what is your point. TS can cache or not, NVE MUST work under both conditions. - 3.3, do you want to cover the distributed GW function? [Lucy] Section 3 is to describe types of tenant system. Getaway is one of them. But it points out that the gateway may be implemented on NVE for some cases, where it is named as distributed GW function on NVE. The distributed GW function on NVE is not a tenant system. The draft specifies the distributed GW function, which is important for the interoperability. - Furthermore, what is the intention of this document? Is it an input document to the architecture draft? Or is it a standard track for NVE function level definition? If the intention of this draft is the NVE function level description, maybe we can work together. I'll send you more comments on the section 4. [Lucy] I state it clearly in the top. Many Thanks, Lucy Have a nice day Zu Qiang >-----Original Message----- >From: [email protected] [mailto:[email protected]] On Behalf Of >Lucy yong >Sent: Tuesday, November 12, 2013 2:43 PM >To: [email protected] >Subject: [nvo3] FW: New Version Notification for >draft-yong-nvo3-nve-01.txt > >Hello NVO3 Community, > >This is a new draft. This document specifies NVE data plane functionality. >These functionality specifications are necessary for the >interoperability between an NVE and its attached tenant systems and >between the NVEs. The data plane functionality described in this >document is independent of NVE or >NVO3 control plane functionality. However the specifications in this >document can support any control plane solution and are helpful in the >control plane protocol development. > >We love to hear people feedbacks on the draft. Some sections in the >draft are not completed yet and will be updated in next version. > >Regards, >Lucy > >-----Original Message----- >From: [email protected] [mailto:[email protected]] >Sent: Tuesday, November 12, 2013 12:24 PM >To: Lucy yong >Subject: New Version Notification for draft-yong-nvo3-nve-01.txt > > >A new version of I-D, draft-yong-nvo3-nve-01.txt has been successfully >submitted by Lucy Yong and posted to the IETF repository. > >Filename: draft-yong-nvo3-nve >Revision: 01 >Title: Network Virtualization Edge (NVE) >Creation date: 2013-11-11 >Group: Individual Submission >Number of pages: 17 >URL: http://www.ietf.org/internet-drafts/draft-yong-nvo3-nve-01.txt >Status: http://datatracker.ietf.org/doc/draft-yong-nvo3-nve >Htmlized: http://tools.ietf.org/html/draft-yong-nvo3-nve-01 >Diff: http://www.ietf.org/rfcdiff?url2=draft-yong-nvo3-nve-01 > >Abstract: > This document specifies Network Virtualization Edge (NVE) data plane > functionality for Network Virtualization Overlays (NVO3). These > functionality specifications are necessary for the interoperability > between an NVE and its attached tenant systems and between the NVEs. > > > > > > >Please note that it may take a couple of minutes from the time of >submission until the htmlized version and diff are available at tools.ietf.org. > >The IETF Secretariat > >_______________________________________________ >nvo3 mailing list >[email protected] >https://www.ietf.org/mailman/listinfo/nvo3 _______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
