Document: Hierarchical Mobile IPv6 Mobility Management Reviewer: Joel M. Halpern Review Date: 19-April-2008 IETF LC End Date: 22-April-2008 IESG Telechat date: 24-April-2008 Summary: Almost ready for publication as a proposed standard. Comments: Significant: As I read this protocol, if a Home Agent is beign used, then packets arriving at the MAP from the Home Agent will be tunneled, with the original packet inside a header from the HA to the MAP. The MAP will then wrap that up in a tunnel and ship it to the LCoA. This means that HMIP nodes that use Home Agents need to be preapred to receive doubly-encapsulated packets. But there is no description of this in the Mobile Node Operation section 6.1. This also seems to imply that for communications where a Home Agent is being used, and bidirectional tunneling is being used with the Home Agent, then the Mobile Node needs to generate doubly-encapsulated packets. The section on detecting MAP failures could use improvement. I realize that this is the same text as was in RFC 4140. However, it seems to allow fairly common undetected failures. It seems to me that stronger warning text about the weaknesses of manual intervention, and the possibility of undetected failures even with some keep-alive protocols ought to be mentioned. Minor: The changes section indicates that Dynamic MAP discovery was removed. However, section 5 indicates that the distance "SHOULD be set to 1 if Dynamic MAP discovery is used." The fourth paragraph of section 6.1 indicates that a Mobile Node must send a binding update to is Home Agent. If it is using a Home Agent, this is correct. However, this spec now indicates that a Home Agent is not mandatory. So this paragraph should say that the Binding Update must be sent if the Home Agent is being used. Given that Home Agents are option, the second paragraph of 6.1.1 should be clarified. Either communicating directly with a CN is only permitted when one has a Home Agent, or the RCoA is permitted in the Home Address option of the Binding Update.