Re: [MEXT] draft-xia-mext-hioptv4-01 and draft-ietf-mip6-hiopt-17 comments
xiayangsong <[email protected]>
| Newsgroups | gmane.ietf.nemo |
|---|---|
| Message-ID | <CB60645E6241144CB82269604373757AE6D72F@dfweml503-mbx.china.huawei.com> |
Hi Tomek Thank you for reviewing the document. Please see my inline comments... BR Frank -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Tomasz Mrugalski Sent: Thursday, March 17, 2011 2:58 PM To: MEXT WG Cc: Behcet Sarikaya Subject: [MEXT] draft-xia-mext-hioptv4-01 and draft-ietf-mip6-hiopt-17 comments Dear MEXT WG, Behcet asked me to review draft-xia-mext-hioptv4-01.txt. It refers to draft-ietf-mip6-hiopt-17.txt, so I'd like to briefly comment on both of them. Disclaimer: I'm not involved in MEXT activity (nor do I intend to). I'm quite familiar with DHCPv6 and to some extent DHCPv4 protocols, so my comments are related to those aspects. draft-ietf-mip6-hiopt-17: MIP6 Relay Agent Option defined in Section 3.2 of draft-ietf-mip6-hiopt-17 is specific to one option only. On the other hand Ted's RSOO proposal is generic and can be applied to any options. This make RSOO better approach. I strongly suggest to modify Section 3.2 of draft-ietf-mip6-hiopt-17 to reference RSOO draft, rather than define its own dedicated option. draft-xia-mext-hioptv4-01: In Section 1 you enumerate all IPv4-IPv6 permutations, without any justification why they are actually needed. What's the use case there? My understanding is that there is Dual Stack mobile node. Is it supposed to do DHCPv4 only? Why? If client visits network that is IPv4 only, how is it supposed to do MIPv6? Frank=>This is all about Dual Stack MIPv6. There is one scenario that a IPv4 only mobile node needs to discover a home agent address through DHCPv4 protocol. According to specification of DSMIPv6, the home address needs both an IPv4 and IPv6 address. If you have dual stack mobile node, it should run both DHCPv4 and DHCPv6 clients, each configuring its own protocol family. Frank=>DSMIPv6 mobile node does not necessarily support both IPv4 and IPv6, or DHCPv4 and DHCPv6. I think Ted's major objection was Section 4.2, where DSMIPv6 Relay Agent Option is defined. Why do you need this option for? What options is it supposed to carry? Are you expecting it to convey DHCPv6 options, as defined in mip6-hiopt-17? That would be wrong on many levels. You would encapsulate DHCPv6 suboptions into DHCPv4 options. DHCPv4 clients typically don't have parsing capabilities of DHCPv6 option formats. Frank=>Just I explained aforementioned, we need define DHCPv4 option for carrying home agents information(such as IPv6 addresses). We can't encapsulate DHCPv6 suboptions into DHCPv4 options because the client is DHCPv4 only. At current form, this draft is unclear at best. Who is including OPTION_DSMIP6_RELAY option? Relay? Frank=>Access routers as DHCP relay and DHCPv4 servers includes the option. My generic recommendation is that authors should should start with: - explaining why you need this instead of using DHCPv4 for IPv4 configuration and DHCPv6 for IPv6 configuration. - redesigning those option to not requiring conveying DHCPv6 options in DHCPv4. - explain your use case and the rationale behind it. Frank=>Thank your recommendation, and we will update this draft accordingly. References: RSOO draft: http://datatracker.ietf.org/doc/draft-ietf-dhc-dhcpv6-relay-supplied-options/ Please keep me on cc: as I'm not subscribed to mext mailing list. -- Tomek _______________________________________________ MEXT mailing list [email protected] https://www.ietf.org/mailman/listinfo/mext