RE: [Mip6] concensus on [issue94] fixing encapsulation mechanisminIPv4 network

["Tsirtsis, George" <tsirtsis at qualcomm.com>]

I prefer A, since I think it is good to allow the mobile to request UDP
tunneling for whatever reason.

I agree with Vijay that it is not clear how the MN should decide to make
such request. Indicating that the MN SHOULD NOT be setting the U flag by
default is a good idea. Another BCP type text could instruct
implementers to try setting the U flag if things are failing with no
apparent reason...e.g., registration apparently works but no data are
ever received... maybe due to a FW on the path.

There is also an almost parallel discussion on whether the HA should be
able to override a request or not. I do not think you can "force" an HA
to do anything...this is why such things are called "requests" and not
"commands". I think the correct thing to do here is to allow the HA to
reject the BU with an error code set to something like "UDP Tunneling
not allowed". It is then up to the MN to try to reregister with U flag
set to "0" or to give up.

Thoughts?
George

> -----Original Message-----
> From: Vijay Devarapalli [mailto:vijay.devarapalli at azairenet.com]
> Sent: Wednesday, March 28, 2007 3:45 PM
> To: Sri Gundavelli
> Cc: mip6 at ietf.org; 'Keiichi SHIMA'
> Subject: Re: [Mip6] concensus on [issue94] fixing encapsulation
> mechanisminIPv4 network
> 
> Sri Gundavelli wrote:
> >
> >
> > -----Original Message-----
> > From: Vijay Devarapalli [mailto:vijay.devarapalli at AzaireNet.com]
> > Sent: Wednesday, March 28, 2007 7:32 AM
> > To: Sri Gundavelli
> > Cc: 'Hesham Soliman'; 'Keiichi SHIMA'; mip6 at ietf.org
> > Subject: Re: [Mip6] concensus on [issue94] fixing encapsulation
> mechanism
> > inIPv4 network
> >
> >>> Reason:
> >> >From the MIP4 experience, many times and for various reasons,
> >>> deployments choose UDP Tunneling even in the absence of NAT. The
> >>> ability for the client to hint that choice from any connected
> >>> network is important.
> >
> >> The problem is that the client may not know. It might be
> >> setting the flag all the time. Do we have a mechanism to
> >> say when the MN should be setting this bit?
> >
> > The preference comes from the type of connected access network
> > and other characteristics local to that network. So, the mobile
> > node is in a better position to make that choice. Off course,
> > the home agent always has the final say, in the form of configured
> > policy.
> 
> So can we then add a sentence saying
> 
>    It is recommended that the mobile node SHOULD NOT set the 'U'
>    flag by default, since using UDP encapsulation when not
>    required introduces an unnecessary overhead for all data traffic.
:)
> 
> My preference is still for 'B' though.
> 
> Vijay
> 
> >
> > But, sure, the logic as when to enable this on the mobile is
> > difficult. Mostly, it might be an operator initiated explicit
> > setting, on a need basis.
> >
> >
> > Sri
> 
> 
> _______________________________________________
> Mip6 mailing list
> Mip6 at ietf.org
> https://www1.ietf.org/mailman/listinfo/mip6

_______________________________________________
Mip6 mailing list
Mip6 at ietf.org
https://www1.ietf.org/mailman/listinfo/mip6