IPv6 - real vs theoretical problems

Tue Jan 11 19:31:43 UTC 2011

On Jan 11, 2011, at 10:45 AM, Michael Loftis wrote:

> On Fri, Jan 7, 2011 at 3:44 PM, Owen DeLong <owen at delong.com> wrote:
> <snip>
>> There are multiple purposes to /48s to residential end users.
>> 
>> DHCP-PD allows a lot of future innovations not yet available.
>> 
>>        Imagine a house where the border router receives a /48
>>        from the ISP and delegates /64s or /60s or whatever to
>>        other routers within the house.
>> 
>>        Each home entertainment cluster may be one group of
>>        networks with its own router.
>> 
>>        The appliance network(s) may have their own router(s).
>> 
>>        RFID tags on groceries may lead to a time when your
>>        home automation server can gather up data from your
>>        refrigerator, pantries, etc. and present the inventory
>>        on your mobile phone while you're at the grocery store.
>>        No more need to maintain a shopping list, just query
>>        the inventory from the store.
>> 
>> These are just the things that could easily be done with the
>> technology we already know about. Imagine what we might
>> think of once we get more used to having prefix abundance.
> <snip>
> 
> Having more address space won't help most of these uses, and as for
> why, take a look at the proposed situation with for example home media

Yes, it will...

> serving/sharing systems by TiVo, Apple, etc. They all require that the
> units be within the same broadcast domain or that there be a
> configured bridge of some sort if they even allow that topology.  They

That is the current state of the art which is the direct result of the lack
of address space and the lack of the features I am describing making
this absolutely necessarily.

> (actually rightfully) assume that the network topology is flat, single
> broadcast domain, and mroe and more use Multicast DNS (which I've seen

Yes, that assumption is valid today. Future technology can change that
assumption in useful and meaningful ways.

> called a bunch of different things)  More to the point, your average
> home user can not technically fathom anything more complicated than
> "plug it in" -- and many begin to fail to set something up properly
> when its extended to something as complicated as "plug it in, push a
> button" or "plug it in, type some numbers into the device"

DHCP-PD will allow for hierarchical topology that is not more complicated
than "plug it in". No button push, no typing something in. Literally plug
it in.
> 
> Your average home user has no reason at all for anything more than a
> PtP to his/her gateway, and a single prefix routed to that gateway.

Correct. I'm just saying that prefix should be a /48 so that the gateway
can work with the other gateways inside the house to designate the
best topology within the house. Note, this is all automated. It doesn't
require the end-user to actually do anything other than plug it in.

> There are most certainly a few (which includes I'm sure 99% of the
> NANOGers!) subscribers who can and will use more space than that, and
> ISPs most definitely should make /48s readily and easily available for
> those customers, but giving each and every customer a /48 (or really,
> even a pair of /64s, one for the PtP, one delegated) is almost
> certainly overkill.  The devices won't use the extra space unless

That is today only thinking. Toss out your IPv4 scarcity-based assumptions
about what is possible. IPv6 does have new features and new capabilities
that we are just beginning to consider.

> there's some automagic way of them communicating the desire to
> eachother, and appropriately configuring themselves, and it would have
> to be very widely accepted.  But there's no technical gain.  A typical

It's called DHCPv6-PD and it already exists. That's the point!!

> household would probably have less than about 50, maybe 100 devices,
> even if we start networking appliances like toasters, hair dryers and
> every single radio, tv, and light switch.
> 
It's not about the number of devices. That's IPv4-think. It's about the number
of segments. I see a world where each home-entertainment cluster would
be a separate segment (today, few things use IP, but, future HE solutions
will include Monitors, Amps, Blu-Ray players, and other Media gateways
that ALL have ethernet ports for control and software update). The
kitchen appliances would probably have their own segment. A refrigerator
or pantry may have a front-end router that separates the household
backbone from the network interfacing all the RFIDs contained within
the device. I'm sure there are other examples where automated
segmentation of the network can, does, and will make sense.

We're just starting to explore this. The point is to have address delegation
policies which don't interfere with this development.

> Just my 2 cents worth.

I'll see your $0.02 and raise you $0.48 ;-)

Owen