Hacker News Reader

Comment by nanijoe | Hacker News Reader

There are a lot of addresses that should be 'reclaimed' . These days, most enterprises really only need a few public addresses, and many of them are 'sitting on' large chunks of addresses they acquired in the 90s.

View on HN

It would be a better use of everybody's energy to actually get to work on the IPv6 transition, rather than trying to squeeze another couple of years out of IPv4. I think it should be pretty much obvious to everyone by now that until we actually hit the wall of IPv4, no one is going to get serious about the transition. NAT is not a great solution - it's a hack, and as anybody that has had to set up a server behind a NAT knows, it can be a real pain.

I'll wager that most residential Internet users are NAT'd, as are most corporate Internet users. The Internet works better today than it did back in 1996, when virtually nobody was NAT'd (back then, the big imposition was dynamic addresses instead of static).

NAT'd by their own router you mean? Because that seems like a world of difference from ISP-level NAT which I can imagine breaking a lot of shit.

When people argue that IPv6 is necessary because NAT is a terrible hack that breaks applications, they are talking about the concept of NAT, the thing that makes it hard to pass a callback address in protocols they design and that requires their routers to peek into their FTP traffic.

I'm sure ISP NAT is at some level worse than home NAT, but the point is, we can scale our IPv4 address space if companies stop, for instance, spending /19's to give every one of their desktops a routable IP address (which are then firewalled off the Internet anyways). We'll simply have to accept some form of self-managed NAT to do that.

I'm fine with that. We're freighting IP with too much responsibility and it's holding us back. We'd have multicast in several different service models by now if people would just stop trying to get Cisco IOS routers to do it for us.

Please don't. This would just be another temporary solution allowing us to get even more invested in a flawed technology. We need to do the move as soon as possible so I really hope no "hero" comes in and buys us 5 more years on this.

As opposed to "reclaiming" addresses i.e. returning addresses to the registries, we should come up with a scheme were institutions can sell their addresses (i.e. IPv4 addresses should become property). This would give institutions that possess very large blocks of addresses a large incentive to effectively utilize their address space.

We could go on for a long-time with IPv4 if the above were to occur.

While I agree, reclaiming them would only delay the inevitable by a few months or a couple years, perhaps.

There certainly is. As of a couple of years ago IBM's 9.0.0.0/8 network was only used for their intranet and not routed outside. I believe several legacy /8's are in the same boat.

IIRC, HP is still sitting on three /8 networks - one from HP, Compaq, and DEC.

HP has two /8 blocks. Many people who at one point owned Class A blocks have given them back over the years. What's crazy, though is that the US Department of Defense still owns eleven /8 blocks.

Reply on news.ycombinator.com