Re: [MEXT] using MR-HA tunnel vs. combining BU // Re: WGLC for draft-ietf-mext-nemo-pd-01.txt

[Ryuji Wakikawa <ryuji.wakikawa at gmail.com>]

Hi Julien and Yungui,

Actually, it's not only 3 round trip and is not optimization problem..
This is a big issue of DHCP-PD.

Let me explain the issue clearly.

According to the NEMO-DHCP-PD spec, before receiving a prefix,
a mobile router must create a tunnel with its HA.

For this purpose, the mobile router needs to act as a RFC3775
compliant mobile host before starting DHCP-PD.  Why? If there is no
prefix, the mobile router cannot sends a NEMO-BU (i.e. R-flag set) to
its Home Agent. If there is no valid prefix for the mobile router, HA
returns BA with "Invalid Prefix" status value which is treated as the
fatal error in RFC3963.

Well, RFC3963 allows for a mobile router to act as a mobile host, but
we didn't assume that a mobile router switches its mode between
RFC3775-mobile-host and RFC3963-mobile-router dynamically.

For instance, RFC3963 said in section 6.2,
"If the Home Agent has a valid binding cache entry for the Mobile
Router, and if the Binding Update has the Mobile Router Flag (R) set
to a value different from that in the existing binding cache entry,
then the Home Agent MUST reject the Binding Update and send a Binding
Acknowledgement with status set to 139 (Registration type change
disallowed). However, if the Binding Update is a de-registration
Binding Update, the Home Agent ignores the value of the Mobile Router
Flag (R)."

At the end, how the boostrapping goes?

1. MR acts as a MN and sends MIP6-BU (Rflag unset) to establish a
   tunnel for DHCP-PD
2. MR starting DHCP-PD
3. After DHCP-PD completion, MR de-registers the RFC3775-binding
4. MR sends NEMO-BU (R-flag finally set) for its mobile network prefix
   acquired by DHCP-PD

I don't think this is right way to go....

I now changed my mind that the use of BU/BA is more straightforward
for NEMO prefix delegation.

ryuji


On 2008/12/11, at 10:50, Julien Laganier wrote:

> We had this discussion for a long time abd it has concluded some time
> ago already: WG consensus is to use DHCP PD.
>
> (w/o questioning the value of optimizing RTTs for a procedure which
> isn't in a critical path, e.g., handover)
>
> --julien
>
> On Thursday 11 December 2008, Yungui Wang wrote:
> > Hello
> >
> > Here is one comment about using MR-HA tunnel for DHCP-PD.
> >
> > In this draft, the MR registration processing needs 3 round trip
> > between MR and HA. i. BU to HA; (getting and binding MR_HoA)
> > ii. DHCPv6 message over MR-HA tunnel. (getting MNP)
> > iii. BU to HA; (binding MNP)
> > While, if PD is combined within BU, it is only 1 round trip.
> >
> > From implementation of viewpoint, the later seems well done prior of
> > the former. Maybe I have lost something, can anyone tell me the story
> > why we gave up the latter in the new version? Thanks.
> >
> > B.R.
> > Yungui
> >
> >
> > ----- Original Message -----
> >  From: marcelo bagnulo braun
> >  To: mext ; Julien Laganier ; Ralph Droms
> >  Sent: Tuesday, November 04, 2008 8:02 PM
> >  Subject: [MEXT] WGLC for draft-ietf-mext-nemo-pd-01.txt
> >
> >
> >  Hi,
> >
> >  We now start the WGLC for:
> >
> >  DHCPv6 Prefix Delegation for NEMO
> >  draft-ietf-mext-nemo-pd-01.txt
> >
> >  http://www.ietf.org/internet-drafts/draft-ietf-mext-nemo-pd-01.txt
> >
> >
> >  Please send comments about the draft till the November 19.
> >
> >  Regards, Julien and marcelo
> >
> >
> >  _______________________________________________
> >  MEXT mailing list
> >  MEXT at ietf.org
> >  https://www.ietf.org/mailman/listinfo/mext
>
>
>
> --
> --julien
>
> [ New email address: julien.laganier.IETF at googlemail.com ]
> _______________________________________________
> MEXT mailing list
> MEXT at ietf.org
> https://www.ietf.org/mailman/listinfo/mext

_______________________________________________
MEXT mailing list
MEXT at ietf.org
https://www.ietf.org/mailman/listinfo/mext