Re:Best IPv6 Read ever (not the article) (Score 1)

A common mistake people make with IPv6 is considering it as only IPv4 with more bits for the address. That is not how the protocol was intended and it will not be used like that. The least significant 64 bits of an IPv6 address are meant only for hosts; the smallest possible subnet you can have in IPv6 consists of 2^64=1,8*10^19 IP addresses. It will never be practicle to have this many devices in a given network segment, in fact, from my experience it is not feasible to have much more than 1000 devices on a given subnet. This results in a great redundancy of the last 64 bits. The idea is that hosts can get a network prefix and then determine their own 64 bits (note that they are not obliged to use their EUI address (in other words the MAC address for most ethernet adapters) for determining the 64 bits). The IPv6 addresses do not replace arp, ICMPv6 does.

Currently, the only unicast IPv6 addresses that are publicly assigned are in the 2000::/3 range (http://www.iana.org/assignments/ipv6-unicast-address-assignments/) - so if you're gonna make jokes about obscure IPv6 addresses, please let them start by 2 or 3 (unless you're talking about multicast/link-local/site-local addresses in which case there are other ranges).

This whole root-servers-going-IPv6 news are not that big news though. More like a milestone on the way to IPv6. In any forseeable future v4 will be used along v6 - the world will probably have destroyed itself before IPv4 vanishes...