Comcast releases 'AFTR': Open Source IPv4 to IPv6 Routing | Hacker News Reader

Comcast releases 'AFTR': Open Source IPv4 to IPv6 Routing

blog.comcast.com

19 pointsbacarter17 comments

Good for them. I have very few kind words for Comcast, RCN, AT&T, and their ilk, but I have to say that this was a surprisingly forward-thinking decision on their part.

The only reason I'm still a Comcast customer is because I want to be in their IPv6 beta trials.

The only reason I'm still a Comcast customer is because there's no other broadband alternative in my area.

No, 128 kbps is not broadband. So, I'm stuck.

This is really cool to see. IPv6 is the internet's "peak oil" problem, or a "long emergency" - most seem to think of the transition as a "maybe someday" nice to have, without realizing how few blocks remain in the IPv4 address space.

It's refreshing to see a major ISP take the initiative to prepare their networks, but also release software to help others route between IPv4 and IPv6 networks. Speaking of which, you can download it here: https://www.isc.org/software/aftr

That said...I'd prefer that most of the source weren't a 13,217-line file with rare, cryptic comments called "aftr.c" ;-).

PDF'd manpages: http://u.phoreo.com/rh.pdf

IPv6 is more like "solar power" in your analogy. We're going to run out of oil/IPv4 someday and have to replace it with something, but not necessarily IPv6.

Actually IPv6 is probably more like "atomic fusion" - some utopian grand project that will fix everything but nobody quite knows how we're going to get it to work.

The idea that IPv4 will be replaced with something that isn't IPv6 is no longer credible. We need a solution in 2012, and nothing else is going to be ready by that time.

Well it wouldn't take much for something else to outpace the adoption rate of IPv6. I think something will come along that is backwards compatible with v4. We're already seeing things like carrier grade NAT that are just bandaids but we can live with bandaids for a long time...

After this long, I'm starting to think we're actually more likely to see cataclysmic earthquakes and flooding in 2012 than adoption of IPv6.

I found this relevant NANOG37 (2006) presentation pretty interesting about Comcast's IP space problem: http://www.nanog.org/meetings/nanog37/abstracts.php?pt=MzQ5J... "Managing 100+ million IP addresses"

On a related note, can I as a consumer do anything to help IPv6 along?

Actually, you can sign up to be a Comcast IPv6 beta tester: http://www.comcast6.net/volunteer.php

I wonder what are the potential problems.

I've a Linux system as a NAT firewall in front of everything else. It's also running an email server and a couple websites (via a dynamic IP provider). Behind it there's a couple laptops, two desktops, game consoles, iPhones, etc.

I wonder what's going to stop working if I switch to IPv6 for my frontend.

Not really. You could enable Teredo so that the fraction of IPv6 traffic on the Internet rises from 0.1% to 0.2%, but nobody is going to take action based on that.

For a long time there was a resistance in the ipv6 crowd from doing something like this, due to the fact that it would make having an ipv4 only service an upside-only proposition (all ipv4 and ipv6 nodes can reach you) thus slowing down the adoption of ipv6.

Since ipv6 can't possibly be adapted slower than it is now shrug

  static char addrbuf[8][NI_MAXHOST]; /* XXX ugly thread-unsafe hack */

  /* XXX: assume this succeeds */
  inet_ntop(family, addr, cp, NI_MAXHOST);

At least they're honest about it. Calling them out because they were blunt in a comment just encourages them to be more subtle, not necessarily write better code.

And particularly in an open-source project, bluntness like that is helpful since it tells other contributors what needs to be replaced ASAP.

I know we all love to hate Comcast, but you can find similar short-term hacks in any code base. 3 lines of code doesn't necessarily make the rest of released source useless.

Reply on news.ycombinator.com