Re: [Geopriv] [geopriv] #38: Section 1

From: Marc Linsner ^lt;mlinsner@cisco.com>
Date: Fri Aug 13 2010 - 12:02:34 EDT

Bernard,

A host using *any* LCP must initiate another message exchange to get updated
location information when they determine it's necessary. This is the same
whether the LCP is HELD, DHCP, LLDP-MED (or wait for 30 sec. for next
packet), or .11(?). Hence, I don't understand your statement (in the
context of this document), "where continuous triggering of DHCP might be
most objectionable". Are not 'bits-on-the-wire', 'bits-on-the-wire'?

My change to the current paragraph below stems from when I read the current
text, one could infer that obtaining updated location information via the
DHCP LCP mechanism is not possible. In fact, it is possible using a
DHCPInform message (yes, renew of a lease also).

Further, I believe it's fairly obvious that *only* using the DHCP RAIO
options as a location determination mechanism renders device level location
(more granular than just AP) *almost* impossible in an 802.11 network (since
RAIO doesn't support measurements). So, when the connection AP is a DHCP
relay agent, I don't believe there are any RAIO options that would allow the
DHCP server to determine, by itself, device level location. Obviously
someone could develop a connection between the DHCP server and the .11
location determination infrastructure, but I don't think we should go there,
nor even mention such in the document.

Hence, I believe the most limiting factor to using DHCP as the LCI for
wireless networks is the impracticality of the DHCP server performing device
level location (vs. AP level location) determination using the available
RAIO information. I also believe that we don't need to point this out as
any developer will realize this within 30 seconds of thinking about.

Does this make any sense?

-Marc-

On 8/13/10 10:46 AM, "Bernard Aboba" <bernard_aboba@hotmail.com> wrote:

> My question with the paragraph is what it is trying to
> convey about the "limited applicability".
> We had discussed how DHCP might work reasonably well in
> static or nomadic wireless scenarios, so that wireless
> itself was not the issue.
>
> The original text seems to suggest that the issue is movement of
> the mobile host without changing the point of attachment (which
> will trigger a DHCP exchange).
>
> My presumption is that this is an issue in situations where
> the location could change a great deal without a change to the point
> of attachment. Unfortunately, it is these situations (e.g. wireless
> WAN networks) where continuous triggering of DHCP might be
> most objectionable.
>
>
>
> -----Original Message-----
> From: Marc Linsner [mailto:mlinsner@cisco.com]
> Sent: Friday, August 13, 2010 6:27 AM
> To: geopriv issue tracker; bernard_aboba@hotmail.com
> Cc: geopriv@ietf.org
> Subject: Re: [Geopriv] [geopriv] #38: Section 1
>
> Bernard,
>
>
> On 8/10/10 5:10 PM, "geopriv issue tracker" <trac@tools.ietf.org> wrote:
>
> In addition to the text that was removed, some text from
>
> draft-thomson-geopriv-3825bis was edited into Section 1:
>
> "The options defined in this document have limited applicability for
>
> mobile hosts. Typically DHCP clients refresh their configuration in
>
> response to changes in interface state or pending lease expirations. As a
> result, when a mobile host changes location without subsequently
>
> completing another DHCP exchange, location configuration information
>
> initially obtained via DHCP could become outdated."
>
> [BA] By replacing the text on wireless usage with a statement of
>
> applicability relating to mobile uses, the overall impression that
> Section
> 1 leaves is a focus on wired uses.
>
>
>
> [ML] After re-reading the above text, I wonder if it might be better
> understood to re-word:
>
> "The options defined in this document have limited applicability for
>
> mobile hosts. Typically DHCP clients refresh their configuration in
>
> response to changes in interface state or pending lease expirations. As a
> result, when a mobile host changes location it MUST complete another DHCP
> exchange to keep location current. Otherwise, the location configuration
> information initially obtained via DHCP could become outdated."
>
> Comments?
>
> -Marc-
>
>
>

_______________________________________________
Geopriv mailing list
Geopriv@ietf.org
https://www.ietf.org/mailman/listinfo/geopriv
Received on Fri, 13 Aug 2010 12:02:34 -0400

This archive was generated by hypermail 2.1.8 : Fri Aug 13 2010 - 12:03:02 EDT