2 ms·
The "ancient" code probably already speaks IPv6 due to whatever method used to hook IP connectivity to its VTAM interfaces. Or due to its age it has multiprotoc
by p_l 3y ago
The "ancient" code probably already speaks IPv6 due to whatever method used to hook IP connectivity to its VTAM interfaces. Or due to its age it has multiprotocol support anyway.
It's the BSD Sockets using code that is a problem - the OSes has supported v6 for a long time now.
- bombcar 3y agoActually routing IPv6 is usually possible with existing equipment, once you figure out all the settings (and potentially avoid the bugs). What usually ends up being the issue is some side piece of software that isn’t ready to log IPv6 addresses or assumptions made about how long they’ll be in some database somewhere, etc. I’ve seen hacks around that, like hashing IPv6 addresses down to 4 bytes so you can store them as fake IPv4 addresses :O
- p_l 3y agoYeah, it's just that it took time to implement, especially since IPv4 ossification (at the time when it was explicitly in works to be removed!) happened partially through introduction of hardware-assisted switching. But we have had fast path switching for IPv6 for I think 20 years now, and it's generally available. Hell, some big networks, the kind where having fast path switching, moved fully over to v6 and simply do not transit v4 (Deutsche Telekom, for example, whose skeleton network is now "flat" IPv6 network and any v4 traffic is encapsulated to 464xlated