Re: [homegate] A new proposed charter (and name) for HOMEGATE

[Mark Townsley <[e-mail]>]

On 9/9/10 4:44 PM, Stephen [kiwin] PALM wrote:
> 
> 
> On 9/9/2010 7:35 AM, Mark Townsley wrote:
> 
>>> Please add "MoCA" (over coax) to the list.
>>
>> Latest version (which is coming to the list shortly, we've been
>> incorporating comments from lots of angles) has "such as" here anyway.
>> I'd rather avoid making the list longer than it already is.
> 
> There are more MoCA deployments than powerline... so the list
> should actually reflect reality.

I think the idea was simply to show diversity, not a popularity contest
by any means. If you think we need MoCA, it's fine with me. Side
questin: Are there any aspects of it that make it difficult to bridge to
other media types (similar to the zigbee requirement that Ralph is
pointing to)?

> 
>>> Seems we should add "bridges" to that list. 8-)
>>
>> Well, hopefully a bridge is transparent to L3 and above. Can we just say
>> this is part of "and more" ?
> 
> Not if the L2 characteristics are different.

But, is there anything specific that homegate would need to do be sure
IP operates through a bridge? Perhaps you are you thinking of QoS here?

> 
>>> Perhaps we should scope that to "personal usage of corporate VPNs"
>>> so we dont stray to far into small/medium office.
>>
>> How about we scope this to "IP connectivity of..."
>>
>> I agree that a corporate VPN is where we start venturing into soho
>> territory. Based on a comment from Lars, we've amended the first
>> sentence to include "home office" ... but not "small office". This is a
>> difficult designation, and perhaps we need to include some text from our
>> earlier exchange to be clear, though I am very sensitive to Andrew's
>> comment that we cannot list all negative cases, and we don't want to end
>> up writing the architecture document in the charter.
> 
> home office is OK.
> 
>> I think that in general, any work we do on IPv4 that also works for IPv6
>> is find and good. But, we very well end up developing solutions that
>> work with IPv6 that do not work with IPv4.
> 
> We cannot ignore the legacy devices in homes

Absolutely IPv6 and IPv4 will have to coexist.

But we can't impose the legacy on IPv6 either. e.g., just because IPv4
in the home looks a like what is described in RFC 5684 today, I don't
think we want to carry that over in the design for IPv6.

While solving something for IPv6 we might very well end up designing
something that would work with IPv4 as well. But, I wouldn't want to
pretend we have to deal with NAPT everywhere in IPv6 just because IPv4
does. This *might* also mean that something is solvable in IPv6 that is
not (realistically) in IPv4.

- Mark

> 
> regards, kiwin
>