6 ms·

Ask HN: What we as community can do to adopt IPv6 faster?

21 points by anthropodie 5y ago

All I can think of is tweeting, emailing big corps or maybe creating a petition, so that these companies take this seriously. Anything else that you guys think maybe more effective?

vorador 5y ago
Genuine question, what do we get out of adopting IPv6? I remember being told for years that we'd run out of IP addresses (which hasn't happened thanks to NAT). What are the advantages of IPv6 from a practical standpoint?
- josephcsible 5y ago
  That you don't have to use NAT, especially carrier grade NAT, anymore.
- nobody9999 5y ago
  >What are the advantages of IPv6 from a practical standpoint? There are a whole raft of advantages: 1. NAT/CGNAT[0] becomes irrelevant (although you can still use local/non-routable prefixes if you wish), eliminating many of the issues around port forwarding and reachability of services; 2. While IPSec is broadly available for IPv4, encryption and authentication are baked into IPv6; 3. The enormous size of the IPv6 addresses would allow every location to have their own 16 bits (or more, if necessary) of routable address space; 4. Despite NAT, we have already run out of IPv4 address space[1] across all regions. There are other good reasons for IPv6 too, but those are the most important IMHO. [0] https://en.wikipedia.org/wiki/Carrier-grade_NAT https://en.wikipedia.org/wiki/Carrier-grade_NAT [1] https://en.wikipedia.org/wiki/IPv4_address_exhaustion https://en.wikipedia.org/wiki/IPv4_address_exhaustion
- vlmutolo 5y ago
  - You pay less money for a single public IP. A single IPv4 costs $40 nowadays [0]. A whole /64 of IPv6 addresses usually comes free when you buy your internet access. - You get extra bits to play with for addressing. Fly.io has done some cool stuff with their "6PN" addresses, built on top of IPv6 and Wireguard. They encode information in the address itself and use eBPF to route on it. - As a result of everyone essentially having infinite public addresses to assign for themselves, we can get rid of NAT, which would enable peer-to-peer apps. See Tailscale's "How NAT Traversal Works" post for a great survey of the techniques used to connect devices behind NATs. (Note that getting rid of NATs does not mean getting rid of private networks or block-all-inbound firewalls. IPv6 allows both of these security measures.) - Some more points here [2] I hadn't seen, particularly around better routing efficiency by getting rid of IP checksums and allowing for smaller routing tables. [0] https://ipv4marketgroup.com/ipv4-pricing/ https://ipv4marketgroup.com/ipv4-pricing/ [1] https://tailscale.com/blog/how-nat-traversal-works/ https://tailscale.com/blog/how-nat-traversal-works/ [2] https://www.networkcomputing.com/networking/six-benefits-ipv6 https://www.networkcomputing.com/networking/six-benefits-ipv...
- DandyDev 5y ago
  We _are_ running out of public IPv4 addresses. I routinely see articles shared on HN about companies buying or selling blocks of IPv4 addresses for increasingly insane prices. I think the market rate is now $40 for a single IP address?
- jimrandomh 5y ago
  IPv4 addresses are easily enumerable, and this is a security problem; people looking for instances of vulnerable software will routinely try every IPv4 address (or search Shodan, a search engine for what-services-are-running-on-each-IP-address). With IPv6, you can't really find random devices that aren't hooked up to DNS or otherwise trying to be found; guessing IPv6 addresses is more akin to guessing passwords.
tomklein 5y ago
I wrote a custom DNS nameserver and a resolver for an upcoming project. I implemented it IPv6 first. I think as more we support and use IPv6, as more it’ll be adopted.
grawlinson 5y ago
My ISP (My Republic; a Singaporean company operating in New Zealand) does not provide IPv6, and I’ve emailed with a request for IPv6. There’s no need for it, apparently.
- geoah 5y ago
  Same with Virgin in the uk.
- ivlad 5y ago
  Tell them you are switching to M1 or StarHub. Both provide IPv6: https://stats.labs.apnic.net/ipv6/SG https://stats.labs.apnic.net/ipv6/SG
yrro 5y ago
I think I'll be dead and buried before my ISP (Virgin Media) rolls out IPv6!
- devplayer0 5y ago
  Which Virgin Media are you on? Here in Ireland they start you with native IPv6 + CGN (you can ask for static IPv4 with a bridge mode). The prefix delegation from the modem is kinda borked though. Also discovered that when I first moved in native IPv6 routing was actually broken in my area (4in6 worked), nothing seemed to get beyond some interior gateways. Took a lot of convincing to get them to actually realise there was a problem and fix it.
  - yrro 5y ago
    UK. I've given up waiting.
- dane-pgp 5y ago
  There's actually a site devoted to tracking the (lack of) progress on this over the past 11 years: https://havevirginmediaenabledipv6yet.co.uk/ https://havevirginmediaenabledipv6yet.co.uk/
geofft 5y ago
Why do you want to adopt IPv6 faster? Personally I think the only realistic move is to advocate for an IPv7, which is backwards-compatible with IPv4 (keep ARP, keep classless addressing, keep DHCPv4, keep seeing NAT as acceptable, don't rely on multicast, etc.) instead of the wacky re-engineering project that IPv6 is. There's a clear need to get more addresses. There's not a clear need for all the other stuff being pushed as part of IPv6.
- taf2 5y ago
  Interesting I’d never heard of ipv7 until now … found this thread https://news.ycombinator.com/item?id=20168718 https://news.ycombinator.com/item?id=20168718
toast0 5y ago
Do companies take tweets, emails or petitions seriously? IPv6 adoption is going to drag along, unless there is a compelling reason for people at large to adopt it. There are reasons, but most of them don't apply to companies that either have enough v4 space or have enough budget to get the space they need. As more consumer networks push towards v6 first/better, you'll see more hosting support v6 as an equal option rather than a second class option, but I don't think we're there yet. A lot of LTE networks do v6 as well or better than v4, but a lot of residential networks still don't do v6 or do it half-ass. A host doesn't lose a lot by being v4only, dualstack adds extra complexity, and v6only loses a lot of potential customers.
- taf2 5y ago
  so as a medium to small business maybe I’ll try pushing on my vendors if they don’t support ipv6 I’ll switch to a competitor that does? Maybe that will help? I already do this with two factor authentication.
speedgoose 5y ago
IPv6 is 25 years old and most companies and people do not care about it. In my opinion it's time for everyone to admit that it has been a failure and move on. The next internet protocol should have stronger arguments, and perhaps start with being interoperable with IPv4.
- marto1 5y ago
  In that spirit even USB didn't get it on the first try. Maybe something along the lines of IPv6 1.1 will do the trick. I'd focus on simplification and killer features.
- deeblering4 5y ago
  Interoperability is really a key thing. The complexity of managing additional ipv6 network bindings, routing, firewalls, service bindings, subnet math, etc, etc makes deployment a huge effort with little real world reward for most. Are there any sites that have fully migrated to ipv6? The sites that I’ve seen running v6 at all are doing so in parallel with v4, which I think just underscores your point.
- rualca 5y ago
  > The next internet protocol should have stronger arguments, and perhaps start with being interoperable with IPv4. Is there any argument that's stronger than having ran out of IP addresses? We already have service providers charging a premium for ipv4 addresses. Isn't that a good enough incentive?
  - speedgoose 5y ago
    We still still have plenty of available ipv4. The current trend seems to be about using what's unused rather than migrating to ipv6. Some ipv4 can be sold thanks to NAT too. I personally would rather pay a cheap premium for a few ipv4 my company needs rather than having to migrate to ipv6. I would most likely still need a few ipv4 anyway.
    - rualca 5y ago
taf2 5y ago
If possible make sure your company websites support ipv6… maybe it’ll be a competitive advantage… I can’t force Verizon to support ipv6 but I can make sure the websites I host or work on do…
rurban 5y ago
We don't want IPv6. We don't need it because we can NAT. We don't want it because we prefer privacy, not being easily tracked all over. Router walls over firewalls. Multicasting didn't work out neither. No common ISP invested into more RAM for their routers just to stream a TV show more efficiently. QoS contradicts the basic internet policies.

21 points by anthropodie 5y ago

All I can think of is tweeting, emailing big corps or maybe creating a petition, so that these companies take this seriously. Anything else that you guys think maybe more effective?

vorador 5y ago

Genuine question, what do we get out of adopting IPv6? I remember being told for years that we'd run out of IP addresses (which hasn't happened thanks to NAT). What are the advantages of IPv6 from a practical standpoint?

josephcsible 5y ago
That you don't have to use NAT, especially carrier grade NAT, anymore.
nobody9999 5y ago
>What are the advantages of IPv6 from a practical standpoint? There are a whole raft of advantages: 1. NAT/CGNAT[0] becomes irrelevant (although you can still use local/non-routable prefixes if you wish), eliminating many of the issues around port forwarding and reachability of services; 2. While IPSec is broadly available for IPv4, encryption and authentication are baked into IPv6; 3. The enormous size of the IPv6 addresses would allow every location to have their own 16 bits (or more, if necessary) of routable address space; 4. Despite NAT, we have already run out of IPv4 address space[1] across all regions. There are other good reasons for IPv6 too, but those are the most important IMHO. [0] https://en.wikipedia.org/wiki/Carrier-grade_NAT https://en.wikipedia.org/wiki/Carrier-grade_NAT [1] https://en.wikipedia.org/wiki/IPv4_address_exhaustion https://en.wikipedia.org/wiki/IPv4_address_exhaustion
vlmutolo 5y ago
- You pay less money for a single public IP. A single IPv4 costs $40 nowadays [0]. A whole /64 of IPv6 addresses usually comes free when you buy your internet access. - You get extra bits to play with for addressing. Fly.io has done some cool stuff with their "6PN" addresses, built on top of IPv6 and Wireguard. They encode information in the address itself and use eBPF to route on it. - As a result of everyone essentially having infinite public addresses to assign for themselves, we can get rid of NAT, which would enable peer-to-peer apps. See Tailscale's "How NAT Traversal Works" post for a great survey of the techniques used to connect devices behind NATs. (Note that getting rid of NATs does not mean getting rid of private networks or block-all-inbound firewalls. IPv6 allows both of these security measures.) - Some more points here [2] I hadn't seen, particularly around better routing efficiency by getting rid of IP checksums and allowing for smaller routing tables. [0] https://ipv4marketgroup.com/ipv4-pricing/ https://ipv4marketgroup.com/ipv4-pricing/ [1] https://tailscale.com/blog/how-nat-traversal-works/ https://tailscale.com/blog/how-nat-traversal-works/ [2] https://www.networkcomputing.com/networking/six-benefits-ipv6 https://www.networkcomputing.com/networking/six-benefits-ipv...
DandyDev 5y ago
We _are_ running out of public IPv4 addresses. I routinely see articles shared on HN about companies buying or selling blocks of IPv4 addresses for increasingly insane prices. I think the market rate is now $40 for a single IP address?
jimrandomh 5y ago
IPv4 addresses are easily enumerable, and this is a security problem; people looking for instances of vulnerable software will routinely try every IPv4 address (or search Shodan, a search engine for what-services-are-running-on-each-IP-address). With IPv6, you can't really find random devices that aren't hooked up to DNS or otherwise trying to be found; guessing IPv6 addresses is more akin to guessing passwords.

tomklein 5y ago

I wrote a custom DNS nameserver and a resolver for an upcoming project. I implemented it IPv6 first. I think as more we support and use IPv6, as more it’ll be adopted.

grawlinson 5y ago

My ISP (My Republic; a Singaporean company operating in New Zealand) does not provide IPv6, and I’ve emailed with a request for IPv6. There’s no need for it, apparently.

geoah 5y ago
Same with Virgin in the uk.
ivlad 5y ago
Tell them you are switching to M1 or StarHub. Both provide IPv6: https://stats.labs.apnic.net/ipv6/SG https://stats.labs.apnic.net/ipv6/SG

yrro 5y ago

I think I'll be dead and buried before my ISP (Virgin Media) rolls out IPv6!

devplayer0 5y ago
Which Virgin Media are you on? Here in Ireland they start you with native IPv6 + CGN (you can ask for static IPv4 with a bridge mode). The prefix delegation from the modem is kinda borked though. Also discovered that when I first moved in native IPv6 routing was actually broken in my area (4in6 worked), nothing seemed to get beyond some interior gateways. Took a lot of convincing to get them to actually realise there was a problem and fix it.
- yrro 5y ago
  UK. I've given up waiting.
dane-pgp 5y ago
There's actually a site devoted to tracking the (lack of) progress on this over the past 11 years: https://havevirginmediaenabledipv6yet.co.uk/ https://havevirginmediaenabledipv6yet.co.uk/

geofft 5y ago

Why do you want to adopt IPv6 faster? Personally I think the only realistic move is to advocate for an IPv7, which is backwards-compatible with IPv4 (keep ARP, keep classless addressing, keep DHCPv4, keep seeing NAT as acceptable, don't rely on multicast, etc.) instead of the wacky re-engineering project that IPv6 is. There's a clear need to get more addresses. There's not a clear need for all the other stuff being pushed as part of IPv6.

taf2 5y ago
Interesting I’d never heard of ipv7 until now … found this thread https://news.ycombinator.com/item?id=20168718 https://news.ycombinator.com/item?id=20168718

toast0 5y ago

Do companies take tweets, emails or petitions seriously? IPv6 adoption is going to drag along, unless there is a compelling reason for people at large to adopt it. There are reasons, but most of them don't apply to companies that either have enough v4 space or have enough budget to get the space they need. As more consumer networks push towards v6 first/better, you'll see more hosting support v6 as an equal option rather than a second class option, but I don't think we're there yet. A lot of LTE networks do v6 as well or better than v4, but a lot of residential networks still don't do v6 or do it half-ass. A host doesn't lose a lot by being v4only, dualstack adds extra complexity, and v6only loses a lot of potential customers.

taf2 5y ago
so as a medium to small business maybe I’ll try pushing on my vendors if they don’t support ipv6 I’ll switch to a competitor that does? Maybe that will help? I already do this with two factor authentication.

speedgoose 5y ago

IPv6 is 25 years old and most companies and people do not care about it. In my opinion it's time for everyone to admit that it has been a failure and move on. The next internet protocol should have stronger arguments, and perhaps start with being interoperable with IPv4.

marto1 5y ago
In that spirit even USB didn't get it on the first try. Maybe something along the lines of IPv6 1.1 will do the trick. I'd focus on simplification and killer features.
deeblering4 5y ago
Interoperability is really a key thing. The complexity of managing additional ipv6 network bindings, routing, firewalls, service bindings, subnet math, etc, etc makes deployment a huge effort with little real world reward for most. Are there any sites that have fully migrated to ipv6? The sites that I’ve seen running v6 at all are doing so in parallel with v4, which I think just underscores your point.
rualca 5y ago
> The next internet protocol should have stronger arguments, and perhaps start with being interoperable with IPv4. Is there any argument that's stronger than having ran out of IP addresses? We already have service providers charging a premium for ipv4 addresses. Isn't that a good enough incentive?
- speedgoose 5y ago
  We still still have plenty of available ipv4. The current trend seems to be about using what's unused rather than migrating to ipv6. Some ipv4 can be sold thanks to NAT too. I personally would rather pay a cheap premium for a few ipv4 my company needs rather than having to migrate to ipv6. I would most likely still need a few ipv4 anyway.
  - rualca 5y ago

taf2 5y ago

If possible make sure your company websites support ipv6… maybe it’ll be a competitive advantage… I can’t force Verizon to support ipv6 but I can make sure the websites I host or work on do…

rurban 5y ago

We don't want IPv6. We don't need it because we can NAT. We don't want it because we prefer privacy, not being easily tracked all over. Router walls over firewalls. Multicasting didn't work out neither. No common ISP invested into more RAM for their routers just to stream a TV show more efficiently. QoS contradicts the basic internet policies.