Using one-byte buffers, one hopes. :) -mel via cell
On Jul 8, 2015, at 8:49 PM, Dave Taht <dave.taht@gmail.com> wrote:
On Wed, Jul 8, 2015 at 7:49 PM, Karl Auer <kauer@biplane.com.au> wrote:
On Wed, 2015-07-08 at 21:03 -0500, Mike Hammett wrote: I wasn't aware that residential users had (intentionally) multiple layers of routing within the home.
No, what they often have is multiple layers of nat. I was at a hotel once that had plugged in 12 APs, serially, wan, to lan, to wan, to lan, to wan ports... because the Internet is a series of tubes, right?
You, we, all of us have to stop using the present to limit the future. What IS should not be used to define what SHOULD BE.
What people NOW HAVE in their homes should not be used to dictate to them what they CAN HAVE in their homes, which is what you do when you provide them only with non-globally-routable address space (IPv4 NAT), or too few subnets (IPv6 /56) to name just two examples.
Multiple layers of routing might not be what is now in the home, but it doesn't take that much imagination to envision a future where there are hundreds, or even thousands of separate networks in the average home, some permanent, some ephemeral, and quite possibly all requiring end-to-end connectivity into the wider Internet. Taking into account just a few current technologies (virtual machines, car networks, personal networks, guest networks, entertainment systems) and fast-forwarding just a few years it's easy to imagine tens of subnets being needed - so it's not much of a leap to hundreds. And if we can already dimly see a future where hundreds might be needed, history tells us that there will probably be applications that need thousands.
Unless of course we decide now that we don't WANT that. Then we should make it hard for it to happen by applying entirely arbitrary brakes like "/48 sounds too big to me, let's make it 1/256th of that."
In my case I have completely abandoned much of the debris of ipv4 and ipv6 - using self assigned /128s and a mesh routing protocol everywhere, giving up on multicast as we knew it, and all I need is one /64 to route my (almost entirely wireless) world.
Somehow I doubt this will become a common option for others, but it sure is easier than navigating the slew of standards, configuring centralized services, and casting and configuring limited and highly dynamic ipv6 subnets around.
Regards, K.
-- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Karl Auer (kauer@biplane.com.au) http://www.biplane.com.au/kauer http://twitter.com/kauer389
GPG fingerprint: 3C41 82BE A9E7 99A1 B931 5AE7 7638 0147 2C3C 2AC4 Old fingerprint: EC67 61E2 C2F6 EB55 884B E129 072B 0AF0 72AA 9882
-- Dave Täht worldwide bufferbloat report: http://www.dslreports.com/speedtest/results/bufferbloat And: What will it take to vastly improve wifi for everyone? https://plus.google.com/u/0/explore/makewififast