click to edit master title style click to add subtitle © 2008 wichorus inc. all rights reserved....

16
Click to edit Master title style Click to add subtitle © 2008 Wichorus Inc. All rights reserved. CONFIDENTIAL - DO NOT DISTRIBUTE rfc3775bis Issues July 28, 2008 [email protected]

Upload: ginger-helena-white

Post on 01-Jan-2016

214 views

Category:

Documents


0 download

TRANSCRIPT

Click to edit Master title style

Click to add subtitle

© 2008 Wichorus Inc. All rights reserved. CONFIDENTIAL - DO NOT DISTRIBUTE

rfc3775bis Issues

July 28, [email protected]

Click to edit Master title styleList of issues

• #1 Last Accepted SQN• #2 Removing DHAAD mechanism• #4 Remove references to site-local addresses• #7 DSMIPv6 BU format and RFC 3775• #8 Application using the care-of address• #9 Simultaneous Mobility• #10 The usage of "HA lifetime"• #11 De-registration when returning home• #12 BErr sent by HA too, not only by CN• #13 Home Link Detection• #14 References to Bootstrapping• #15 BRR sent by HA too, not only by CN• #16 HA behaviour upon MN returning Home

Click to edit Master title styleMinor note/suggestion about issue description

• Document owner (me) “owns” the result of resolving the issue (buck stops here)

• Issue owner “responsible” for checking to see if I have correctly summarized the issue and the discussion from the mailing list

• “defect” – something wrong with the document

• “enhancement” – something that should be done to improve the document

• Suggestions for best use of issue tracker welcome!

Click to edit Master title style

Issue #1: Last Accepted SQN (defect)[Ahmad Muhanna <[email protected]>]

Current specification requires the mobile node to discard Binding Acknowledgement (BA) with a sequence number out of range

Problem: Then there’s no way for the mobile node to get resynchronized

Proposal: Mandate mobile node to use the supplied sequence number in authenticated BA (Status 135)

Click to edit Master title style

Issue #2: Removing DHAAD mechanism (defect) [Jean-Michel Combes <[email protected]>

Suggestion: Now that we have RFC 5026, expunge DHAAD

• 80 email messages, many points of disagreement

• Is not important for PS submission of rfc3775bis

• Proposed modification to

Click to edit Master title style

Issue #4: Remove site-local addresses (defect)[George Tsirtsis <[email protected]>]

Problem: site-local addresses have been deprecated, but rfc3775bis still makes reference to them

Solution: change site-local addresses to ULAs

This seems almost like a no-brainer, but then again maybe people are also dissatisfied with ULAs.

Proposal: just make the simple changes for PS, unless ULAs also become deprecated

Click to edit Master title style

Issue #7: DSMIPv6 BU and RFC 3775 (enhancement) [Tero Kauppinen [email protected]]

Problem: IPv4 address in encapsulating header can never match the source address (IPv6 CoA) of BU

• Proposal 1: Require an IPv4 mapped IPv6 address in the BU header

• Proposal 2: view DSMIPv6 as a document updating the behavior specified in rfc3375bis, punt

Proposal 2 seems to have better support (?)

Click to edit Master title style

Issue #8: Application using the CoA (enhancement)[Julien Laganier <[email protected]>]

Observation: after RFC 3775 was published, the IETF has subsequently published RFC 5014 to enable a mobile application to use the care-of address as a source address

Proposal: cite RFC 5014 as informational

Click to edit Master title style

Issue #9: Simultaneous Mobility (defect) [Ashutosh Dutta <[email protected]>

Problem: signaling for RO can fail if mobile node and correspondent node move simultaneously

• Proposal: fall back to reverse tunneling – but, when?Right away?After one try?After MAX_RETRIES?Retry in parallel with reverse tunneling?

Click to edit Master title style

Issue #10: The usage of "HA lifetime" (defect)[Ryuji Wakikawa <[email protected]>]

Problem: does MN expunge home agent address when its lifetime expires? Does this on whether the HA address was statically configured?

• Question: is there a default value for the HA address lifetime?

• Answer 1: a matter for local configuration

• Answer 2: expunge the HA address

Click to edit Master title style

Issue #11: De-registration returning home (defect)[Vijay Devarapalli <[email protected]>

When returning home, mobile device MUST deregister

• Why was the original text “SHOULD”?

• But what if the mobile does not want to use the home link? (e.g., multi-homed device) (then, it didn’t actually “return home”??)

• Or, perhaps we ought to keep the “SHOULD” to encourage future innovation?

• Or, perhaps add text explaining why “SHOULD”?

Click to edit Master title style

Issue #12: BErr sent by HA too (defect)[Alexandru Petrescu <[email protected]>

Is it possible for HA to send BErr?

• For unrecognized mobility headers

Seeming consensus on easy text change

Click to edit Master title style

Issue #13: Home Link Detection (defect) [Suresh Krishnan <[email protected]>

Problem: need to specify how mobile node detects whether it has returned to its home network

• Proposal: add text from draft-krishnan-mext-hld

Click to edit Master title style

Issue #14: References to Bootstrapping (enhancement)[Vijay Devarapalli [email protected]]

Problem: base document was written before there were specifications available for dynamic assignment of a home agent to a mobile node

• Proposal: add informative citation for RFC 5026, ??draft-ietf-mip6-bootstrapping-integrated-dhc-05

Click to edit Master title style

Issue #15: BRR sent by HA too (enhancement)[Ahmad Muhanna <[email protected]>]

Should HA be allowed to send BRR?

• Proposal: it could increase reliability of MN registration

• Objection: This could be considered new functionality and thus not appropriate for rfc3775bis

Click to edit Master title style

Issue #16: HA behaviour when MN returns Home (defect)[Pascal Thubert <[email protected]>]

Problem: if mobile node reboots, it may not remember to reclaim control of its home address when it detects it is on the home network.

• Proposal: let MN use DAD or oDAD, and update existing local neighbor caches by setting ‘O’ bit.

• Concern: maintain backwards compatibility