Dear authors, I have reviewed draft-ietf-bess-virtual-subnet-05 as part of the OPS-DIR directorate reviews and you should treat all comments as any IETF Last Call comments. I found the document well written and easy to read/comprehend. The document is about ready for publications with small nits. I did not find any text about operational aspects of the virtual subnet. I was not really even expecting to see those but if there _are_ already known good management / operational practises specifically in the virtual subnet context those would be nice to have listed. Few specific areas what I was thinking are where the line is when having more than one default gateway gets justified or are there operational concerns related the number of host routes etc. Not requiring this text though if the authors think it is not needed. Small nits: * VRF, VDP, TTL are never expanded. * Section 3.7. line 422: "..contain ARP/ND entries of local hosts As a result, .." ^^^^^^^^ I have issues parsing this sentence. * Since this is Informational RFC all references could also be Informational. Clarifications and semi-substantial: * Section 4.2. Line 485 on non-support of IP Broadcast and Link-Local Multicast. Since IPv6 ND also relies on link-local multicast for quite a few things this section is a bit confusing. Are you saying IPv6 ND is also handled using L2VPN? I do not get that implression from the rest of the document. You might want to open this specific part a bit more how IPv6 ND is handled, since you have proxying also in place. * Section 7. security considerations.. I find the statement a bit bold that there are no additional security risks. At least the I would expect to see a reference to RFC6583 when it comes to IPv6 ND and e.g., opeartion of PE routers / ND proxies. - Jouni