What the hell is… IPv6?
Why it's happening, what it means, who's doing it
There are just 4 billion possible IP addresses if we do not move from version 4 of the Internet Protocol to version 6. After the move, there would be 340 trillion trillion trillion (3.4 x 10**36) possible addresses under v6, because of the 128-bit address space. Scalability is just one reason to make the move, but the new levels of security that could be achieved - quite apart from much better network management, especially for mobile devices - make the move highly desirable. A further benefit of switching to IPv6 would be to get better and more reliable performance. Users mostly have little awareness of the routing problems in the global backbone, and the associated quality of service issues, but they are real enough to make it desirable to accelerate the move to v6. There are those who suggest that IPv4 extensions are capable of fixing problems as they arise, but introducing techniques like network address translators puts off the day when significantly better security could be achieved. An extremely important security feature is authentication of packets from a host. IPv4 servers cannot determine if the packets have been received from an acceptable node. A partial answer has been firewalls, but the consequences include a performance hit and restrictive policies. A blind eye has been turned towards the presence of sniffers for market research and network traffic analysis, but there is a great deal of latitude for misuse at this level, with resultant security and confidentiality compromises. With v6, there is end-to-end encryption at the network layer, and further handshaking protocols, to ensure a serious level of security unobtainable with v4. Apart from work on routers, the changes to accommodate IPv6 mostly require the inclusion of the protocol in the client operating system, with transparency so far as users are concerned. In the non-Microsoft world, in nearly all cases operating system developers are well-advanced in the incorporation of v6, although there are provisions in place to allow v4 and v6 to co-exist for many years. So far as implementations of IPv6 on non-Microsoft platforms is concerned, some of the horses have completed the course: IBM's AIX has it built in, as does Novell's NetWare, and BSDI. Some Linux distributions have it. Sun has a prototype for Solaris 7 (and first made v6 available in 1995); Compaq has a Tru64 prototype available for download; and HP has a developers' kit. Horses currently running include Apple, now building IPv6 into MacOS X, and some Linux distributors who have not yet included it. Microsoft is galloping in the first furlong, with the jockey valiantly shouting: "Wait for us, we're the innovators." ®
Sponsored: Hyper-scale data management