[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Home network topology



I would be very careful in making sweeping statements about home networks.
I have four "home" networks that come quickly to mind. Two are the  
networks in my daughter's and one of my son's apartments; they each  
use a typical low end router with combined 10/100 Ethernet and  
802.11g wireless capabilities; they wire things that stand still and  
use wireless for things that move around. The third is my eldest  
son's fraternity house, which is best described as a piece of work.  
The fourth is my own home, a map-of-sorts for which is at
    ftp://ftpeng.cisco.com/fred/v6ops/Fred_home_network.pdf

The reason for the internal addressing behind Linksys routers in that configuration has to do with the configuration of those routers and the bugs in our Nokia and Samsung phones, which only seem to do the 802.11 hotspot service correctly if the routers are in a firewall configuration. Apart from the special configuration for low rate beaconing, which results in longer battery life on the phones, I would happily replace the Linksys routers with access points.
The distinction between wired and wireless uses is important with the  
industry trends in the direction of the "connected home". While my  
marketing people tell me that everyone that installs a router in  
their home uses wireless for everything and can be expected to try to  
run HDTV over it along with everything else, my kids (who are not  
networking experts in any sense) pragmatically reply that they have  
tried it and it doesn't work. Those of us who understand the  
pragmatic aspects of networking could write treatises on why, but the  
fact is that we don't have to. Picking one, when I moved my UCSD son  
to an apartment last summer, he specified to me that he wanted  
wireless in the living room, but for the five student desks he wanted  
wired service (we installed 10/100/1000 switches). He wanted the  
desks wired because that configuration had worked well in his prior  
apartment, and wireless didn't. As I say, in general, wiring things  
that stand still (like TVs and people's desks) and using wireless for  
more flexible purposes trades off signaling characteristics in a  
pretty sensible way, and sensible people do so.
In general, I would encourage the folks who manage address space to  
enable different options for different users depending on their needs  
and willingness to pay; I might find useful options for /64, /60,  
and /56 addressing for a typical home, SOHO, or SMB. /48 is  
appropriate for larger installations, and for truly large companies  
(my own, with 60,000 employees each of whom has a remote corporate  
LAN in their home in addition to labs and workspaces at work, comes  
to mind), even an addressing plan that gives 65,000 routable prefixes  
might be too small if there are more than 65,000 separately routable  
endpoints. The /44 available in a ULA makes sense in that context.
In short, one size doesn't fit all, and giving folks the ability to  
do sensible things for a budget trade-off makes sense.
On Mar 12, 2008, at 8:41 AM, james woodyatt wrote:

On Mar 11, 2008, at 15:48, Rémi Denis-Courmont wrote:
For instance, 6to4rd is now proposing that only a /64 be assigned  
to a given
customer. If we expect the home to be fully bridged, that's fine.  
Otherwise,
there's going to be an interesting problem.
I expect that most home networks for the foreseeable future will be  
well-served by just a single bridged 802.1 network.  A significant,  
and non-ignorable fraction of them will need to run an interior  
routing protocol behind the residential Internet gateway.  One  
scenario is the remote worker whose office network is required by  
policy [or tax law] to be firewalled from their personal network.   
Other potentially more common scenarios spring to mind, but sadly,  
I'm unable to offer a disquisition on them at this time.

--
james woodyatt <jhw@apple.com>
member of technical staff, communications engineering