[support] About the architecture of the MIPv6 daemon [Second Part]

[Romain KUNTZ]

Hi,

On 2009/03/09, at 18:14, Angel Bartomeu Bonillo wrote:
> Can any of the MCoA fellows add more details about this issue?

The algorithm for MCoA is roughly the same than the one Arnaud  
described, and it is applied to each interface registered in the  
configuration file.

MCoA however restrict handovers to horizontal ones only (i.e. no  
handovers are performed from one interface to another) : when one  
interface/address/network is not available anymore, a de-registering  
BU is sent via another interface to remove the associated binding at  
the HA.

The src/movement.c source file will also give you better details on  
this.

Cheers,
romain


> 2009/3/9 Arnaud Ebalard <arno at natisbad.org>
> Hi,
>
> Angel Bartomeu Bonillo <angelbartomeu at gmail.com> writes:
>
> > First of all, I need to thank to those who replied to my first
> > question (*Arnaud
> > Ebalard,** **manish Jamwal and **Romain KUNTZ)* on the thread at
> > http://ml.nautilus6.org/pipermail/support/2009-February/ 
> 000616.html . I had
> > problems with my email and I was forced to register again with  
> this other
> > address since I was somehow unable to receive any message from the  
> list.
> > Hence I will try to continue the thread I once started from here.
> >
> > As for being more specific about my needs (Answering to Arnaud  
> Ebalard's
> > question), I am building a mobile node for a vehicular network  
> that must be
> > able to interact with different routing protocols that will be  
> running in
> > the mobile node. The node may have Internet access through  
> different access
> > networks i.e. UMTS, GPRS, WIFI an so on. So my intention is to use  
> NEPL to
> > manage the multiple CoA once acquired on the different links  
> available.
> >
> > I have been going around the code of the mip6d (NEMO and MCoA  
> patched) and
> > though I have found many interesting things but, could anybody  
> tell me based
> > on which events the daemon make the decision of starting a  
> handover and
> > destroying the tunnel in one interface to redirect the traffic to  
> another
> > tunnel on other interface?.
> >
> > I have found so far that one of these events is the status of the
> > interfaces. As I have seen the code register a listener by netlink  
> to get
> > the changes in the interfaces. Does the code take any other thing  
> into
> > account, like router advertisement timer out or something like that?
>
> In UMIP, the decision to start a handover is mostly due to  
> notification
> of a change in link/address configuration received via Netlink, for
> instance when a link becomes available (or is not anymore available)  
> or
> an address is added or deleted on a given interface (or an interface
> disappear). It may also happen due to an asynchronous event, for
> instance if some address lifetime (CoA) becomes null or a router
> lifetime become null (IIRC). Then, based on available addresses,
> interfaces, and associated preferences, the decision is taken to  
> perform
> a handover.
>
> I don't know if the MCoA patch changes anything to the logic described
> above.
>
> Cheers,
>
> a+
> _______________________________________________
> Support mailing list
> Support at ml.nautilus6.org
> http://ml.nautilus6.org/mailman/listinfo/support
>
>
>
> -- 
> Ángel Bartomeu Bonillo.
> C/ Federico García Lorca 28º Esc1 Bajo  B
> 30009 Murcia (España)
> _______________________________________________
> Support mailing list
> Support at ml.nautilus6.org
> http://ml.nautilus6.org/mailman/listinfo/support