Good. One smaller editing change in the document: s/a NVE/an NVE/
Lucy From: LASSERRE, MARC (MARC) [mailto:[email protected]] Sent: Thursday, September 19, 2013 12:53 PM To: Lucy yong; Pat Thaler; Benson Schliesser; [email protected] Subject: RE: [nvo3] WGLC for draft-ietf-nvo3-framework-03 Lucy, This was implied but I will make this small editing change. Thanks, Marc ________________________________ From: Lucy yong [mailto:[email protected]] Sent: Thursday, September 19, 2013 6:13 PM To: LASSERRE, MARC (MARC); Pat Thaler; Benson Schliesser; [email protected]<mailto:[email protected]> Subject: RE: [nvo3] WGLC for draft-ietf-nvo3-framework-03 Marc, If we want to keep the framework in a high-level view, IMO, we at least need the following statement in section 2.3. An NVE can provide L2 service, L3 service, or both to a tenant network and support multiple tenant networks. Lucy From: LASSERRE, MARC (MARC) [mailto:[email protected]] Sent: Thursday, September 19, 2013 5:10 AM To: Lucy yong; Pat Thaler; Benson Schliesser; [email protected]<mailto:[email protected]> Subject: RE: [nvo3] WGLC for draft-ietf-nvo3-framework-03 Lucy, Section 2.3 does state that NVEs can support both L2 and L3 services. Marc ________________________________ From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Lucy yong Sent: Wednesday, September 18, 2013 10:04 PM To: Pat Thaler; Benson Schliesser; [email protected]<mailto:[email protected]> Subject: Re: [nvo3] WGLC for draft-ietf-nvo3-framework-03 Here is another point. The goal of the draft is to provide a framework for Data Center Network Virtualization. Today, it is very common that the hosts in a data center may belong to the same subnet and different subnets. Bridging the hosts in the same subnet and routing them across different subnets are common networking practice. It seems that the Data Center Network Virtualization in industry does not intend to change this practice either (see VMware NSX). However, current draft describes the framework for L2 VN overlay and L3 VN overlay, respectively, which is not clear to me that address a virtual network attached by the hosts that may belong to the same subnet and different subnets. Not clear to me if the L3VN overlay is sufficient to cover this case. I brought this up before and like to bring it up again. An NVE can provide L2 and L3 service for a tenant system that attaches to a VN. If the tenant system sends a packet to a destination that on the same subnet, it sends to that destination MAC directly and NVE forwards on the dMAC too. If the tenant system needs to send a packet to the destination on a different subnet, it sends the packet to the default router w/ default MAC. NVE can be the default router. In this case, receiving the packets w/ NVE MAC as dMAC, NVE performs IP lookup to get a tunnel EP as well as dMAC and destination TS VN ID. In L3 VN overlay model, NVE does not need to have default MAC address, no need to resolve the mapping between dest. IP and dMAC/VNID. Should we address this in the framework? Regards, Lucy From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Pat Thaler Sent: Friday, September 13, 2013 2:17 PM To: Benson Schliesser; [email protected]<mailto:[email protected]> Subject: Re: [nvo3] WGLC for draft-ietf-nvo3-framework-03 The framework is in pretty good shape but not quite ready. I am concerned that the framework is fuzzy on the relationship of hypervisors to tenant systems and NVEs. This became apparent during recent discussions on the security draft. I looked at the framework and the control plane drafts to see what they said before posting. The figures in the framework don't show the hypervisor - TS just connects to the NVE when in cases where the TS is a VM the hypervisor would be between them. The figures in the hypervisor NVE draft show the hypervisor as containing the VMs but they don't show the TS. "Hypervisor" only appears in the terminology section in the definition of NVE (which says the NVE could be part of a hypervisor virtual switch), in its own definition, and in the definitions of server and virtual switch. Clearly the role of the hypervisor in the framework should be described beyond that. At least, its role in communicating about TSes to the NVE and controlling the network behavior of TSes that it hosts (e.g. filtering). Regards, Pat From: [email protected]<mailto:[email protected]> [mailto:[email protected]] On Behalf Of Benson Schliesser Sent: Thursday, September 12, 2013 8:08 PM To: [email protected]<mailto:[email protected]> Subject: [nvo3] WGLC for draft-ietf-nvo3-framework-03 This email begins a one week working group last call for draft-ietf-nvo3-framework-03. Please review the draft and post any comments to the NVO3 list. This working group last call will end on Friday 20-September-2013. Cheers, -Benson & Matthew
_______________________________________________ nvo3 mailing list [email protected] https://www.ietf.org/mailman/listinfo/nvo3
