RE: [dhcwg] WG last calls on several WG documents

["Bernie Volz \(volz\)" <volz at cisco.com>]

Ted:

It doesn't leave out discovering prefix delegations. When a requestor
queries for an address, either the stateful address lease *OR* the
delegated prefix that "contains" the queried address is returned,
depending on what is "leased".

Yes, there is no specific query to ask whether a specific Prefix (prefix
+ prefix-length) is leased, but I didn't see that that was needed and
the requestor might not even know what it is (otherwise, why would it be
querying).

Remember that the basic model is to support relay agents (such as
CMTS's) that have lost "state". The relay agent will attempt to recover
state when a packet arrives with a source or destination address
(depending on the direction of the traffic) that the router does not
have in its tables. As the router only has an address, it would not
query for a delegated prefix (and that would perhaps require TWO queries
-- one for the address and another for a delegated prefix).

So, by querying for the address the relay agent would get back the
client and lease information for either the stateful address or the
delegated prefix per the draft.

We can certainly add a query by prefix if that is desired in the future,
but I just find it unlikely that the relay agent would know what the
prefix length is in advance.

- Bernie

-----Original Message-----
From: Ted Lemon [mailto:Ted.Lemon at nominum.com] 
Sent: Friday, January 12, 2007 5:37 PM
To: Stig Venaas
Cc: dhcwg
Subject: Re: [dhcwg] WG last calls on several WG documents

On Jan 10, 2007, at 7:43 AM, Stig Venaas wrote:
> Please note that Ralph asked this to verify consensus
> for starting last calls. Based on the positive input we
> got for agentopt-delegate and srsn-option we started a
> last call for those. Once that is completed we will
> see if we now also got sufficient support for starting
> last calls on leasequery and ero.

I'm mildly uncomfortable with leasequery simply because it's a rather  
complex draft and I haven't had the time to sit down and try to  
implement it, so I'm worried that there are bugs in it.   Also, it  
seems to leave out discovering Prefix Delegations, which seems like a  
significant omission.   But with those caveats, I'm okay with it going  
forward to the next step, as long as we understand that it may need  
fixes if errors are discovered during implementation.   Possibly those  
can be folded in when the solution for Prefix Delegation is described.

I'm also fine with ERO going to last call.



_______________________________________________
dhcwg mailing list
dhcwg at ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg

_______________________________________________
dhcwg mailing list
dhcwg at ietf.org
https://www1.ietf.org/mailman/listinfo/dhcwg