[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [MEXT] Review of Re: I-D Action:draft-ietf-monami6-multiplecoa-12.txt
Hi George, Christian, Benjamin, and all,
Let me first reply you the confusion of the binding de-registration operation.
In the MCoA doc, there are several scenarios of binding de-registrations.
I tried to summarize how MCoA document expects to be in each scenario.
Hope this clarify your concerns.
The mobile node is away, and has the three registrations in the home
agent's binding cache
v6HoA BID1 -> v6CoA1
v6HoA BID2 -> v6CoA2
v6HoA BID3 -> v6CoA3
1. Removing a binding from a foreign link
MN removes a binding (BID2) from the foreign link
+ sending de-registration BU from the foreign link
- lifetime=0
- BID-option: BID2, No-flag (no care-of address should be given)
2. Returning home completely. (Same as RFC3775)
MN uses the interface attached to the home link only.
+ sending de-registration BU from the home link
The de-registration BU is same as RFC3775's one.
- lifetime = 0
- No BID-options (no care-of address should be given)
3. Use the simultaneous home and foreign link (Normal MCoA behavior)
MN returns home with the interface (BID2) and continues using both interfaces.
+ sending de-registration BU from the home link
- lifetime = 0
- BID-option: BID2, H-bit=1 (no care-of address should be given)
- the Mobility Header Link-layer Address Option
4. Use the simultaneous home and foreign link (Flow binding specific behavior)
Two approaches to create a home binding.
(1) sending BU from the home link
- lifetime > 0
- BID-option: BID2, H-bit=1 (no care-of address should be given)
or
(2) sending bulk-BU from the foreign link
- lifetime > 0
- BID-option:
BID1, H-bit=1, (CoA is given)
BID2, H-bit=1, (HoA is given as a CoA)
BID3, H-bit=1, (CoA is given)
5. Removing multiple bindings at once (Replacements)
MN removes multiple binding (BID2,3) registered for the home address
with the new ones (ex. BID1).
+ sending registration BU from foreign link
- lifetime > 0
- BID-option: BID1, O-bit=1, (CoA is given)
# Note1: O-bit is originally introduced for bootstrapping operation,
# but it can be used for multiple binding de-registrations.
# Note2: Bulk-BU has similar feature, but it only does add-on.
# Bulk-BU cannot be used for replacement or multiple binding removal.
6. Giving up using MCoA
MN wants to use the RFC-3775 binding instead of MCoA binding.
+ sending registration BU from foreign link
- lifetime > 0
- No BID-option
#Note1: if MN sends BU (lifetime > 0, BID-option: BID1, O-flag, CoA),
#HA creates one binding but with the BID assignment.
#Note2: if MN wants to switch from RFC-3775 binding to MCoA binding,
# it MUST de-register RFC-3775 binding first and then start
registering MCOA bindings.
regards,
ryuji
2009/3/18 Benjamin Lim <benjamin.limck at sg.panasonic.com>:
> Hi Christian,
>
> You are right. I forgot to consider the case where HA is not the
> default router. The HA would be the one to decide to do ND-proxy or not
> based on the topology.
>
> Regards,
> Benjamin Lim
>
> Christian Kaas-Petersen wrote:
>> The H-bit=1 is not used to turn off neighbor discovery at HA.
>>
>> The H-bit=1 is used by the mobile node to tell the home agent
>> it has simultaneous connectivity on home link and one or more
>> foreign links. Then the home agent decides who should do neighbor
>> discovery on the home link (depending on whether
>> the home agent is the only router on the home link, ie topology a
>> of Figure 3, or there are other routers on the home link, ie topology b
>> of Figure 3), and the home agent reports its decision in the status
>> value of the binding acknowledgement. In topology a the mobile node
>> will do neighbor discovery; in topology b the home agent will continue
>> doing neighbor discovery.
>>
>> Christian
>>
>> --- Den ons 18/3/09 skrev Benjamin Lim <benjamin.limck at sg.panasonic.com>:
>>
>>> Fra: Benjamin Lim <benjamin.limck at sg.panasonic.com>
>>> Emne: Re: [MEXT] Review of Re: I-D Action:draft-ietf-monami6-multiplecoa-12.txt
>>> Til: "Christian Kaas-Petersen" <kaaspetersen at yahoo.dk>
>>> Cc: "George Tsirtsis" <tsirtsis at googlemail.com>, "mext" <mext at ietf.org>
>>> Dato: onsdag 18. marts 2009 08.17
>>> Hi Christian, all,
>>>
>>> Pls see inline for some of my comments.
>>>
>>> Christian Kaas-Petersen wrote:
>>>> The packets I proposed I get from draft 12, section
>>> 5.4 second paragraph
>>>> (which details a de-registration with lifetime=0 and
>>> absence of
>>>> care-of address) and section 5.6.3 bullets number 1
>>> and 2 (which detail
>>>> the presence of the H-bit and the absence of the
>>> care-of address).
>>>> To me, there is no need to register a binding where
>>> CoA=HoA (use of
>>>> home link), unless flow binding is used.
>>> [Ben] Agree, being simultaneous connected on a home link
>>> and foreign
>>> link does not imply that flow binding must be used.
>>>
>>> For example, MN wants to receive packets from devices on
>>> MN's home link
>>> through MN's home attached interface. Rather than
>>> having those devices
>>> send the packet to HA who will then forward it to MN's
>>> foreign attached
>>> interface, MN uses the 'H' flag to turn off ND-proxy at HA
>>> and advertise
>>> MN's home attached interface. This way, devices in
>>> the home link will
>>> send packets to MN's home attached interface. On the
>>> other hand, any
>>> packets not from the home link (e.g. from the global
>>> Internet) will
>>> arrive at HA, who in turn forward it to MN's foreign
>>> attached interface.
>>>
>>
>>
>>
>> Find din nye laptop på kelkoo.dk. Se de gode tilbud her - http://dk.yahoo.com/r/pat/mm
>>
>>
>
>
> _______________________________________________
> MEXT mailing list
> MEXT at ietf.org
> https://www.ietf.org/mailman/listinfo/mext
>