Re: [Geopriv] Naive question: LCP == LbyR?

From: Hannes Tschofenig ^lt;Hannes.Tschofenig@gmx.net>
Date: Sat Mar 10 2007 - 14:46:58 EST

Hi Richard,

you raised a very good question and I have a short story for you.

Other SDOs look at the location-based services in a different fashion by
assuming that the network understands the application. Consider the 3GPP
IMS emergency service architecture as one more detailed example.

You need a protocol to fetch the location of an end host. The
identification of the end host for location determination might use a
variety of different identifiers, as we all know (MAC address, IP
address, etc. see draft-ietf-geopriv-l7-lcp-ps-00.txt).

The end host provides these identifier via, for example, SIP signaling
to the respective nodes that perform the lookup. No new identifier needs
to be "invented" there since the existing onces do the job, namely the
identifiers are resolvable to the location of the end host within this
given realm.

Now, we have assumed a different architecture where the entity that
needs todo the lookup is from a different identifier realm (i.e., it
cannot resolve the existing identifiers to the location of the end
host). (Note that by realm I mean that an identifier has a specific
validity domain and semantic within this domain. We phrased this quite
fuzzy, for example, by saying that the LIS is in the access network and
somehow knows how to determine the location of the end host even though
we encountered cases where this is not true.)

What did we do?

We came up with a different identifier that has a much larger validity
domain, a SIP URI or an HTTP URI. This is not a big thing but quite
important. This identifier that is used as a replacement of other
identifiers that have a limited scope allows everyone on the Internet
(with some assumptions behind it) to perform the lookup and to determine
the location of the end host.

Ironically, if you demand authorization for the resolution process you
almost destroy all benefits of the newly introduced new identifier since
you artificially re-create the relationship between the entity that is
able to perform the resolution and the entity that knows where the end
host is.

So, how does this all relate to your question regarding LCP == LbyR?

Ignoring some terminology aspects, and comparing LCP that retrieves an
LO based on IP address/MAC address/etc. and a protocol that uses an HTTP
reference to retrieve an LO one might notice that they do very similar,
if not the same things (ignoring the scope of the identifier validity).

Does this make sense?

Ciao
Hannes

Richard Barnes wrote:
> I've got a simple question, that's basically captured by the subject
> line: What's the difference between a Location Configuration Protocol
> and an L-by-R Dereference Protocol? Compare two use cases:
>
> LCP:
> Step 1. Discover LIS
> Step 2. Send query to LIS
> Step 3. Receive LO / URL
>
> L-by-R:
> Step 1. Receive location URL
> Step 2. Send query to dereference server
> Step 3. Receive LO / URL
>
> I realize that with an LCP, there's a additional need to identify the
> target, but it seems just as easy to embed this into a URI or
> dereference request as to use another protocol mechanism. And it
> seems like using a single protocol for both would save us all some
> time and effort.
>
> Cheers,
> --Richard
>
>
> _______________________________________________
> Geopriv mailing list
> Geopriv@ietf.org
> https://www1.ietf.org/mailman/listinfo/geopriv
>

_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www1.ietf.org/mailman/listinfo/geopriv
Received on Sat, 10 Mar 2007 21:46:58 +0200

This archive was generated by hypermail 2.1.8 : Sat Mar 10 2007 - 14:51:31 EST