Re: [homegate] Proposed Charter Update - 2010/09/09

[Jari Arkko <[e-mail]>]

For what it is worth, I do share many of the same concerns David has. 
After showing the charter to a number of experts, all came back with 
question marks on what the group actually intends to work on. Here are 
some of the questions that the current charter text raises for me:

> * Routed home - automatic prefix assignment and distribution of other
>   configuration information in a network with multiple routers, subnets
>   and data link types.
>   

Does this include the development of new routing protocols, DHCP PD 
extensions, operational advice on how to use existing mechanisms, or 
standardized templates for structuring the site's address space? Are we 
talking about 1-5 routers, or aiming for a more general purpose 
automatic prefix assignment mechanism?

> * Simple naming for entities in the home network, so they can be
>   accessed from within the home network as well as from elsewhere on the
>   Internet.
>   

Does this mean pointing to DNS or possibly to DNS and mDNS as the 
solutions and just providing operational advice on how to run these 
types of networks and what the default configs should be? Or are you 
also planning to extend some of the naming mechanisms to go beyond one 
subnet?

Why are local and global naming bundled here? I presume we have dyndns 
and other solutions for the global naming problem already, and that 
problem space seems different from the local routed network problem.

> * Support for connections to multiple Internet service providers or
>   closed networks (VPNs, walled gardens) at the same time.
>   

This seems like an undesirable overlap with the MIF working group.

> * Effective queuing and forwarding mechanisms to enable efficient
>   communication between network segments of different speeds, latencies
>   and error rates.
>   

I'm not sure what this is, beyond recommending the proper queuing 
algorithms. I certainly hope developing some new forwarding semantics is 
not in scope. But is ECN, BEHAVE NAT specifications etc in scope? And 
again, is the group making a profile of existing specifications or 
collecting them together, or developing some new (e.g., a new queuing 
algorithm).

I am not sure slicing and dicing the deliverables at this stage is the 
right answer to making the group's focus clearer. I for one would like 
to get some answers to the fundamental questions about which of these 
questions are in scope. I wouldn't mind if the eventual document was 
still just one deliverable.

Jari