RE: New (-02) version of IPv6 CPE Router draft is available for review

["Hemant Singh (shemant)" <shemant@cisco.com>]

>> On one hand you say "managed by the ISP" and in the same sentence you
say "modem's software that can automatically configure ... the CPE
router".
>>  ^^^^^                                   ^^^^^^^^^             ^^^^^^

>> My head hurts...

Sorry.  Here is Tylenol for your head. It is a MUST in CableLabs (CL)
eRouter CM (cable modem) standard that after the CM has initialized its
cable interface and acquired an IA_PD then the CM MUST configure the LAN
interfaces. 

>> If cable standards already define some part of CPE router behavior,
then what gaps is this draft trying to fill in?

Here are the gaps:

1. CL only specifies an embedded CPE Router.  We are defining a
standalone CPE Router.  There are enough difference between the two. Of
course, CL standard doesn't discuss cascading of routers since one does
not cascade a modem behind another modem.  The standalone router does
need to discuss cascading. The eRouter doesn't need to deal with LAN
init first mode but a standalone router does. The standalone router is
usually NOT managed at all by the SP.  There are numerous other gaps.

2. We are gathering requirements from the DSL world to also define a
standalone and embedded CPE Router suitable for DSL deployment.

3. We don't agree with certain behavior in the CL embedded Router IPv6
and that is why we present our view.

Some of the other reason we wrote this draft are as follows:

1.  We didn't want DSL going off on their own defining a CPE Router and
then break something for cable because cable already has a embedded
router standard defined. 

2. IETF v6ops was discussing security ad nauseam for a home CPE Router,
but didn't have a specification for a CPE Router.  

We have already seen enough evidence in v6ops that folks call all
problems and issues as Broadband issues when they clearly mean to
discuss only DSL.  For example, Cable doesn't need any link model draft
that came out recently that says it's a Broadband model. Yeah, right!
Call it your DSL model.  We know what our cable link, data models are.
We don't need the emails on Broadband/SOHO provisioning - cable has a
working provisioning systems, aggregator, and modem for IPv6 that are
cable standards compliant for IPv6.  


>> Regardless of whether the modem were a DSL or cable or WIMAX modem,
etc., if there's no connection to the SP on startup then the WAN
interface is 
>> basically not useable for anything until it does have connectivity to
the SP.  I think it's desirable that the behavior and desired end-state
of the >> CPE router should be determinate and not depend on the order
in which the interfaces acquire external connectivity.

Totally agreed.  Don't you see our LAN init first before WAN init define
deterministic behavior for a standalone CPE Router?  Our WAN init first
model is also deterministic.  

We'll continue to work on DSL requirements and make some sensible
behavior out of them for the CPE Router.

Hemant 

-----Original Message-----
From: Antonio Querubin [mailto:tony@lava.net] 
Sent: Monday, July 21, 2008 9:21 AM
To: Hemant Singh (shemant)
Cc: Alain Durand; Brian E Carpenter; v6ops@ops.ietf.org; Wes Beebee
(wbeebee)
Subject: RE: New (-02) version of IPv6 CPE Router draft is available for
review

On Mon, 21 Jul 2008, Hemant Singh (shemant) wrote:

> No.  See my earlier email to you that said when the CPE Router is 
> embedded in, say, a cable modem, then the WAN interface of the CPE 
> Router becomes an internal logical interface (as opposed to a 
> physically available port in a standalone CPE Router).  The logical 
> interface bridges to the cable modem.

No argument there.

> The cable modem is definitely managed by the ISP

Perhaps this is true in some cable broadband networks but in my DSL
world the DSL modem is un-managed by us and in most cases not by the
customer either.  Sometimes it's purchased by the end-customer,
sometimes by us and shipped to the end-customer and it may have some
local management control if it's got an integral NAPT router but
regardless we do NOT manage it.

> and it is this modem's software that can automatically configure the 
> LAN
> interface(s) of the CPE Router.  Cable standards have specified this 
> automatic configuration by the modem.

On one hand you say "managed by the ISP" and in the same sentence you
say "modem's software that can automatically configure ... the CPE
router".
  ^^^^^                                   ^^^^^^^^^             ^^^^^^ 
My head hurts...

> The CPE Router when embedded in a cable modem has to comply to Cable 
> standards.

If cable standards already define some part of CPE router behaviour,
then what gaps is this draft trying to fill in?

> When the WAN interface is internal to the modem there is no question 
> of useable or not.  The interface is not even visible to the user.

Regardless of whether the modem were a DSL or cable or WIMAX modem,
etc., if there's no connection to the SP on startup then the WAN
interface is basically not useable for anything until it does have
connectivity to the SP.  I think it's desirable that the behaviour and
desired end-state of the CPE router should be determinate and not depend
on the order in which the interfaces acquire external connectivity.

Antonio Querubin
whois:  AQ7-ARIN