On 22-okt-2007, at 11:32, Wojciech Dec (wdec) wrote:
Well then, let's hear what the DSL Forum has to say about the
issues surrounding upgrading existing DHCP implementation in
host OSes, which, as far as I can tell, can only be done by
their vendors (i.e., Microsoft and Apple), and how to address
IPv6, both with and without
IPv4 and with and without DHCPv6.
Are you seeking to ask the DSLF to design the protocol or answer on
behalf of Microsoft?
No, what I want to know is how they expect to deal with the situation
where a user has a computer that she wants to connect to the service
and that computer is running (for instance) Windows and the Windows
DHCP implementation doesn't support the new options. Installing a new
DHCP client doesn't doesn't seem a workable option here, so what is
the user expected to do in this situation?
Seriously, I believe that the DSLF has come to the IETF seeking
progress
on a solution that conceptually "fits the bill" (ie the requirements)
from the SDO perceived as owning the (DHCP) protocol definition.
So far, what I see is a lot of Cisco people and maybe one or two
others pushing really hard for this while pretty much everyone else
sees significant issues. The IETF has also developed a technology for
exactly this purpose which seems to be rejected out of hand.
So far, nobody has addressed the IPv6 issue. I don't think the IETF
can go forward with any solution that only works for IPv4.
Please note that DHCP has a long history of vendor extensions. It's
probably easier for everyone involved to go that route rather than to
push the IETF to do work that is severely problematic.
I think the DSL Forum should consider the issues that have been
brought up in this discussion and decide what it wants to do based on
that. Rough consensus that this is a great idea and the dhc wg should
go ahead writing a spec doesn't seem to be forthcoming.
Further note: In the DSLF the target devices for incorporating a
subscriber authentication mechanism ones that are provided by the
SP (eg
video STB's, DSL Routers), and here the stacks are much more
controllable. PC End hosts, while not not being totally ruled out, are
of secondary importance.
Either they are supported or they aren't. There's no middle ground.
The way I read the requirements, they are supported.
That said, AFAIK how a DHCP based solution will
be implemented is still TBD, and so saying that it will require a full
change of stack is premature.
The nature of the change is immaterial, the issue is that the DHCP
client isn't a user serviceable part in a commercial operating system
so the vendor will have to apply any changes, however minor. This
tends to take a long time.
_______________________________________________
Int-area mailing list
[email protected]
https://www1.ietf.org/mailman/listinfo/int-area
