Re: tunneling [Was: Agenda for Vienna]

[marcelo bagnulo <marcelo@it.uc3m.es>]

Hi, 


On Tue, 2003-05-20 at 23:57, Iljitsch van Beijnum wrote:
> On dinsdag, mei 20, 2003, at 22:28 Europe/Amsterdam, Tony Li wrote:
> 
> > Meta note: if we're trying to talk architecture, I'd say that
> > dealing with the identifier/locator relationship is secondary.
> 
> I don't think we can make informed decisions about the architectural 
> approach unless we know how well each approach is going to work out. 
> That's why I keep going into detail.
> 
> > Our primary discussion should be about alternatives to
> > "identifier/locator split" solutions.  That's clearly one
> > architecture, what are others that folks find credible?
> 
> Ok, let me list some solutions that have come up:
> 
> - mobility
> - HIP
> - source routing / conditional source routing
> - controlled deaggregation
> - backup path over second ISP using routing or tunnels
> - address agile transports
> - loc/id
> - geo aggregation
> 

to my understanding multiple of the above solutions are based on the
loc/id separation

That is:
- mobility
- HIP
- address agile transports
- loc/id

Are solutions based on the separation of loc/id roles. I mean all of
them use a fixed identifier for application layer and multiple locators
for routing (in this case in the end -hosts)

Moreover, i would say that all (host) multi-address solutions are based
on loc/id split concept when they attempt to preserve established
connections

- backup path over second ISP using routing or tunnels
This may also be considered a loc/id separation, sine an alternative locator is used to reach a final destination 
(so the original dest address plays the role of a locator)

However, i think that there is some fuzziness in this, so perhaps it would be interesting to have 
a more precise definition about what we are talking about when we use the loc/id split expression
I've tried but couldn't find any short definition that i liked...
Perhaps after we have a more precise definition we could ask for opinions and reach consensus more easily.  

So solutions not based on loc/id separation remaining are:

- controlled deaggregation
- geo aggregation

FWIW, i prefer the loc/id split approach (if this means what i think it does)

Regards, marcelo



> I think the limitations of each have been discussed in the past. Some 
> kind of more formal recap would probably be useful at this point. Kurt, 
> is that what you want in your first proposed Vienna meeting or do you 
> only want "active" solutions discussed there?
> 
> > Tunneling has certainly been mentioned.  I've always considered
> > this simply a mechanism for virtualizing the topology.  There's
> > effectively still a single address per host, so there's nothing
> > new in addressing.
> 
> After some discussions on the ipv6mh mailinglist we reached the 
> conclusion that rewriting the original source and destination addresses 
> with something else at the source and then replacing the original 
> addresses before the packet is processed at the destination is simply 
> tunneling optimized for header size at the cost of some state. So I 
> guess that makes the loc/id thing a dynamic version of tunneling...
-- 
marcelo bagnulo <marcelo@it.uc3m.es>
uc3m