[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

FW: Evaluation: draft-ietf-dhc-subscriber-id-06.txt for PS

Is this response acceptable?

Thanks, Bert

-----Original Message-----
From: Thomas Narten [mailto:narten@us.ibm.com]
Sent: Thursday, August 19, 2004 17:04
To: Wijnen, Bert (Bert)
Cc: Margaret Wasserman (E-mail); Iesg (E-mail)
Subject: Re: Evaluation: draft-ietf-dhc-subscriber-id-06.txt for PS 


> An AAA-doctor reviewed this and has a question:

>   This document looks good. One question: There seems to be a
>   growing number of identifiers (client-id etc) for users in
>   the DHCP space. Is there some rule set to determine which of
>   the attributes and in which order are used when, say,
>   determining if the same or different IP address should be
>   handed to the client? For instance, if the client-id has
>   changed but the subscriber ID stays the same, what do you
>   do? Or is this all left to policy?

> Do we have an answer?

There are sometimes a lot of different ways of specifying
(essentially) the same things in DHC. Some of that is history, due to
poorly specified options leading to ambiguity and then a desire to
define a more clear way to fix the problem. Thus, there are multiple
options sometimes where a newer one supercedes the older one, but both
are included (by the client) since it isn't known to the client what
the server will support.

The individual documents are supposed to make it clear what the
priority is when multiple options (of overlapping info) are present in
a DHC message.

My sense is that these sorts of identifiers are treated mostly as
opaque strings from the DHC perspective. The client  often has to be
configured to send the string, and the server is configured to process
it. Thus,  it ends up being handled by policy in terms of how the
server actually parses/processes the data.

I'm not sure that there is really a problem with this. I.e., the DHC
model is that the server is smart enough that it can be configured (by
the admin) to do the right thing based on various combinations of
options.

Thomas