IP address space is becoming scarce under IPv4, the main Internet communications protocol. Growing demand for Internet-enabled devices requires more space for growth. Wireless carriers would like to tie a unique IP address to every phone, pager and PDA, and household appliance makers are experimenting with smart connected devices such as refrigerators and washing machines.

IPv6 offers an enhanced addressing scheme that leaves room for growth. IPv6 has been around since the early 1990s, but the lack of a killer application has slowed its acceptance in the commercial world. Recent events indicate IPv6 might take off soon.

Late 2004 saw the activation of Cernet2 - the next-generation Internet in China and the largest IPv6 network in the world. The US government's deadline for its IPv6 roll-out and compliance is February 2008. Businesses, especially those with Asia-Pacific or US government interests, might find themselves pulled into IPv6 by the need to connect to their colleagues.

IPv6 is a direct plug-and-play replacement for IPv4. All major operating system and network hardware vendors support IPv6, and ISPs are starting to offer IPv6 connectivity. The only thing remaining that holds back IPv6 deployment is the massive effort required for corporations to re-address millions of computers, routers and other Internet devices. But this effort is slowly starting to gain momentum.

A simpler header
The basic IPv6 header is far simpler than the IPv4 header. Each header contains source and destination addresses for packet, payload length, hop limit (equivalent to the IPv4 Time To Live field) and a field indicating the protocol encapsulated in the packet, such as TCP or User Datagram Protocol (UDP).

Significant differences arise in the size of the address space and its layout. An IPv6 address has four times as many bits as an IPv4 address - 128 vs 32. To simplify using IPv6 addresses, the dot-separated eight-bit decimal fields of IPv4 are replaced with colon-separated 16-bit hexadecimal fields, such as 3ffe:501:185b:1:2e0:18ff:fea8:16f5.

IPv4 addresses are assigned by Regional Internet Registries (RIRs) world-wide, such as the American Registry for Internet Numbers (ARIN) in the Western Hemisphere, using Classless Inter-Domain Routing (CIDR) blocks, such as 134.151.0.0/16. CIDR block assignments range from hundreds to millions of addresses per block, and the numbering of the blocks has little relation to the world-wide address structure.

IPv6 addresses, also assigned by RIRs, are more structured. Various addressing schemes are defined and identified by the high-order bits of the address block. The most popular scheme splits addresses in half -- 64 bits for the network and 64 bits for each device. The high-order 64 bits are composed of 32 bits for the RIR, such as ARIN; 16 bits for the local Internet registry or ISP; and 16 bits for the site to which the address belongs. Each site, reminiscent of what was once called a Class-B address block, allows for up to 65,536 devices.

The low-order 64 bits of a given address are used for the interface identifier, such as a specific interface on a device. In the above example, 3ffe:501:185b:1 represents the specific device and 2e0:18ff:fea8:16f5 represents an interface on that device.

Migration to IPv6 requires the systematic renumbering of all of a corporation's IP addresses, ideally first in small well-structured pilot projects to understand how best to use the address space. This would be followed by renumbering of particular locations or business units with IPv4/IPv6 mapping devices on the boundaries. Eventually, as the entire industry migrates, these mapping devices can be removed from internal boundaries When there's enough of an IPv6 Internet to connect, for example, to all of the given corporation's customers, and vendors are also accessible via IPv6, the external mapping devices could be removed, as well.

Siil is chief architect of Lumeta.