Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I remember the first time I looked at my ip6 address and found my MAC in the host portion. Obvious and predictable as it might be to some folks, it surprised and annoyed me. With a unique MAC in every assigned ip6 address, it's pretty easy to imagine the tracking potential, e.g., a unique ID following a user where ever an ip6 address is assigned. Apparently Apple and even Network-Manager(Linux) automatically randomize/obscure the MAC for ip6 addresses - I don't know about Windows though. Using wicd in Linux, I must either randomize the MAC manually, or configure sysctl.conf accordingly. Otherwise, it's great, I suppose, that every atom in the known universe can have its own ip.


When my residential ISP started allocating IPv6 blocks, they published our postal codes in the whois database. In Canada, in a city, that postcode is basically a dozen buildings (one side of a street block), at least where I live. For a stalker, it's practically a full address.

We pressured the ISP to remove this information. They resisted, said it was policy, we read the RFCs and argued back. A few weeks later it was removed.

Similarly, most Linux distributions now enable IPv6 privacy extensions. Once people settle in their habits, it becomes more difficult to change this type of behaviour.

IPv6 is a big change. Adopt early, influence policy while you can :)


When I was a student, my rDNS was the student housing building number + apartment number. Makes it easy for people to direct abuse requests at least.


That was an issue for early IPv6 implementations. In most OS now, even link-local IPv6 addresses do not necessarily include MAC addresses.

However, public IPv6 addresses are obviously part of provisioned ranges. And without NAT, there's no ambiguity about assignment to devices. So in practice, with IPv6 you're sharing something as unique as a MAC with all peers.

One solution is using proxies with anonymously provisioned IPv6. Tor will eventually handle that. It's also possible for VPN services. FrootVPN, for example, does that already, whether you're connecting to their servers via IPv4 or IPv6.


Most OSes already support some form of using temporary random IPv6 addresses within your provisioned range, including so far as for each individual outgoing communication. Windows has supported it since Vista, IIRC, and defaulted to it since 7, I believe.

It's definitely still a problem if your ISP assigns you too small of an IPv6 range, and doesn't stop people from tracking you as your range, but it does as well, if not better, than NAT at stopping people tracking your individual devices.


To this I've added a DHCP script that randomly selects one of the /64's inside the /60 that my ISP allocates me, it runs on binding a lease after a given fraction of the DHCP lease time. Combined with privacy extensions, this keeps my address fairly unpredictable.


> Most OSes already support some form of using temporary random IPv6 addresses within your provisioned range, including so far as for each individual outgoing communication. ...

Yes. I'm planning to run a Tor exit relay that does something like that. It'll connect with directory servers and upstream relays on a stable IPv4 address, however. Changing the exit IPv6 will screw up bandwidth measurements. But it doesn't matter, because I don't want the relay used as anything but exit role.


windows has two ipv6 addresses on every capable interface by default* : a constant one and a temporary one. neither is based on the MAC AFAIK.

[ * ] caveats apply




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: