Ted, all,

Please see a few additional comments inline. 

[snip]
Okay, so I asked you a specific question about this, and outlined a solution 
that involves resolving the FQDN in the DHCP server rather than in the client.  
 You haven't said why this won't work for you.   Could you speak to that?

CJ: capitalizing on VoIP SBC operation experience, the de-correlation of DNS 
and DHCP functions facilitates redirection operations (e.g. towards the CGN 
that maintains static entries for a given customer), and the opportunity to 
apply similar operational procedures to NAT capabilities whether they are 
embedded in a CGN or in a SBC is something that is often valued by service 
providers, at least from an OPEX standpoint.

[snip]
I explained this early, but I'll reiterate.   If you have two different ways in 
DHCP to configure a value in the client, then you create an interoperability 
problem.   The DHCP client requests one or the other option; if it requests the 
wrong option, the server will not respond, because it wasn't configured with 
that option.

In order to correct this problem, the DHCP server needs to know that the FQDN 
option is an alias for the IP address option.  This requires custom code on the 
DHCP server, just for your option.   Imagine if every single working group that 
wanted a DHCP option wanted custom code for the option as well.   You would 
have more customization code in the server than protocol code.   This is what I 
am trying to avoid by having this discussion with you.

I am not objecting to sending the FQDN, if that is what you really need.   But 
I *am* asking you (that is, the softwires working group) to make a decision: do 
you want FQDN, or do you want IP address.   We had a pretty unequivocal "no" to 
DNS from David Hankins.   We have a pretty unequivocal "yes" to DNS from you.  
I'm asking you to work that out.

CJ: it seems to me that the above comment mostly reflects an 
implementation/configuration-driven concern. I can understand this position, 
but I think IETF WGs are chartered to document tools (namely protocols) that 
are meant to address specific issues and requirements. WGs are not supposed to 
document how to *implement* (let alone configure) such tools, which is 
basically the responsibility of service providers and other players. 

I think the requirement for the FQDN option makes sense (at least in some 
providers' environments), and this is why the standardization of this option is 
useful. This does not mean that this option must be used by each and every 
service provider.

As long as this FQDN information can be made available by means of a DHCP 
option, then it's up to the service providers to deal with corresponding yet 
required configuration tasks.

This is why I don't see any harm in standardizing this option along with the IP 
address option, and I therefore support the need to re-include this option in 
the core spec.

Cheers,

Christian.
*********************************
This message and any attachments (the "message") are confidential and intended 
solely for the addressees. 
Any unauthorised use or dissemination is prohibited.
Messages are susceptible to alteration. 
France Telecom Group shall not be liable for the message if altered, changed or 
falsified.
If you are not the intended addressee of this message, please cancel it 
immediately and inform the sender.
********************************

_______________________________________________
Softwires mailing list
[email protected]
https://www.ietf.org/mailman/listinfo/softwires

Reply via email to