IPv6 enabled Christmas Tree • Hacker News Board

Re: IPv6 enabled Christmas Tree

#21

Post by api_or_ipa » Thu, Dec 15, 2016, 6:58 PM UTC

dsl wrote: post #18

nabaffyub wrote: post #12

Earlier quoted context omitted.

We need to 1) force owners of legacy /8s to give up on them, and 2) make sure all mobile ISPs worldwide and residential connections in developing countries have their customers behind a CGNAT 2) would also help a lot with the upcoming IoT security shitstorm

You do realize most of the legacy /8 owners are actively using them, right?

I'm sure Apple, Ford and the US postal service are all using all 16 million addresses. The fact is that most intra-network traffic doesn't need an IPv4 address since it'll be NAT'd anyways. So even if Apple has 16 million IoT devices around campus, it doesn't actually need 16 million addresses.

Fun fact, Stanford used to have an /8 but returned it since it wasn't using all it's addresses.

Re: IPv6 enabled Christmas Tree

#22

Post by sp332 » Thu, Dec 15, 2016, 7:27 PM UTC

Zikes wrote: post #15

So that's like 16,581,375 IPv6 addresses right? How much would all of those cost?

IPv6 officially keeps the last 64 bits for the "host" part of the address, which means that unless you are abusing the protocol, each subnet gets at least 2^64 addresses. Current "best practice" is for consumer ISPs to give each end-user at least a /48, that is 2^16 = 65,536 possible subnets' worth or 2^80 total address. Your mileage is going to vary depending on your ISP though.

Edit: looks like the RFC has been updated to recommend smaller allocations where appropriate. Still keeps /64 as the smallest allowed subnet though. https://tools.ietf.org/html/rfc6177#page-4

Edit2: fixed devices -> subnets.

Re: IPv6 enabled Christmas Tree

#23

Post by jdcarter » Thu, Dec 15, 2016, 7:29 PM UTC

djsumdog wrote: post #11

I think I'll wait a few days when everyone forgets about this, so I can actually see the effects of my pings .. if it's still up.

I sent some yellow (f0:ff:0) pings, they definitely showed up.

Re: IPv6 enabled Christmas Tree

#24

Post by tlrobinson » Thu, Dec 15, 2016, 7:37 PM UTC

rroriz wrote: post #16

nabaffyub wrote: post #12

Earlier quoted context omitted.

We need to 1) force owners of legacy /8s to give up on them, and 2) make sure all mobile ISPs worldwide and residential connections in developing countries have their customers behind a CGNAT 2) would also help a lot with the upcoming IoT security shitstorm

CGNAT prevents users from hosting personal services - its a PitA.

Yeah, I just discovered my ISP (WebPass) has exhausted their IPv4 allocation [1] and is transitioning residential customers to private IPv4 addresses, and I'm not really sure how to set up a VPN on IPv4 now. I guess I need to set up a tunnel over IPv6 somehow?

1. https://webpass.net/blog/ipv4-exhausted

Re: IPv6 enabled Christmas Tree

#25

Post by a_t48 » Thu, Dec 15, 2016, 7:40 PM UTC

nine_k wrote: post #5

Playing music automatically once a page loads is not the best idea. Please avoid doing so when possible. Even in Christmas season.

It's part of the YT video, I believe.

Re: IPv6 enabled Christmas Tree

#26

Post by grahamedgecombe » Thu, Dec 15, 2016, 7:41 PM UTC

sp332 wrote: post #22

Zikes wrote: post #15

So that's like 16,581,375 IPv6 addresses right? How much would all of those cost?

IPv6 officially keeps the last 64 bits for the "host" part of the address, which means that unless you are abusing the protocol, each subnet gets at least 2^64 addresses. Current "best practice" is for consumer ISPs to give each end-user at least a /48, that is 2^16 = 65,536 possible subnets' worth or 2^80 total address. Your mileage is going to vary depending on your ISP though. Edit: looks like the RFC has been upd…

> each device gets 2^64 addresses

Each subnet gets 2^64 addresses. You can have multiple devices in the same subnet.

Re: IPv6 enabled Christmas Tree

#27

Post by cmurf » Thu, Dec 15, 2016, 8:51 PM UTC

And now I'm convinced we're off the rails and should just revert to being an agrarian society.

Re: IPv6 enabled Christmas Tree

#28

Post by cesarb » Thu, Dec 15, 2016, 9:09 PM UTC

Eagerly awaiting a writeup explaining how it was done. It seems to be an Arduino with a pair of shields, the lower one being an Ethernet shield and the upper one perhaps a protoboard shield?

Re: IPv6 enabled Christmas Tree

#29

Post by syncsynchalt » Thu, Dec 15, 2016, 9:21 PM UTC

nabaffyub wrote: post #12

djsumdog wrote: post #10

Earlier quoted context omitted.

IPv6 needed to happen a decade ago. We have exhausted the IPv4 space. I feel IPv6 suffered from huge usability issues, although it can be worked around with good provisioning systems that automatically setup DNS entries. We're going to get to the point where spinning up a VPS with an IPv4 address is going to get increasingly more and more expensive, unless we make some huge strides in moving to IPv6.

We need to 1) force owners of legacy /8s to give up on them, and 2) make sure all mobile ISPs worldwide and residential connections in developing countries have their customers behind a CGNAT 2) would also help a lot with the upcoming IoT security shitstorm

Each /8 only gives us about a month of runway. Not worth the bother.

Re: IPv6 enabled Christmas Tree

#30

Post by syncsynchalt » Thu, Dec 15, 2016, 9:23 PM UTC

rroriz wrote: post #16

nabaffyub wrote: post #12

Earlier quoted context omitted.

We need to 1) force owners of legacy /8s to give up on them, and 2) make sure all mobile ISPs worldwide and residential connections in developing countries have their customers behind a CGNAT 2) would also help a lot with the upcoming IoT security shitstorm

CGNAT prevents users from hosting personal services - its a PitA.

I (hope to) see CGNAT as the stick - with v6 being the carrot that allows point-to-point apps and personal hosting to work reliably again.

That requires the ISP to offer v6, of course. I'm on Comcast which, say what you will about them, has been years ahead of everyone else for carrier-grade v6 support.