IPv6 adoption vs IPv4

So far IPv6 has managed to grab only a

small part of the Internet pie, but there's plenty of action happening in the

background to make the existing Internet infrastructure ready for it. Will the

Internet ever shift to the new protocol anytime soon? Let's take a look.

The original intention for designing IPv6 was that the

available IP address range in IPv4 was running out quickly. There were just not

enough IP addresses in IPv4 to give to everybody. In fact, even the distribution

of IPv4 addresses globally was skewed, with a major portion of them allotted to

the US and Europe and a small part to Asia Pacific countries. With Internet

connectivity growing rapidly in these countries, this skew will have its effect.

Even otherwise, the distribution is inefficient with many companies sitting with

lots of unused blocks of IP addresses. IPv4 countered both these problems for

more IP addresses by devising Network Address Translation or NAT technology.

This needs no introduction, as most organizations already use it to allow

machines on their network to share an IP address and access the Internet. But

NATing has only slowed down the adoption of IPv6 for some time. It's not going

to stop it completely.

The thirst for more IP addresses is much bigger and

increasing. Consider the mobile revolution that's currently happening

globally.  So many digital devices

like cellphones, PDAs, smartphones, etc have become available today. Most of

these can connect to the Internet, and therefore need an IP address. There's

no way that IPv4 would be able to handle this onslaught of new devices. NATing

would only work to some extent.

More IP addresses isn't the only reason to shift to IPv6,

though that was the original intention. Besides this reason, even the older IPv4

has some other limitations that IPv6 promises to overcome. These can be best

understood by looking at the benefits of IPv6. One is that the new protocol is

supposed to have better routing techniques. Unlike IPv4 implementations, in

which the routers maintain huge routing tables, IPv6 would be able to manage

with smaller tables. That's because a lot of information about the source and

destination is built into the IPv6 header itself. This will allow IPv6 packets

to reach their destinations in fewer hops.

Another limitation that IPv6 overcomes is getting an IP address from a

DHCP server. In IPv4, if a machine is configured to obtain an IP from a DHCP

server, it won't be able to get an IP address if the DHCP server isn't

available. IPv6 on the other hand, is supposed to be smart enough to figure out

an IP address from a routing table or from IP addresses of nearby clients.

Likewise, IPv6 is supposed to have better support for security, with integrated

IPSec support. Even if we look at the addressing scheme, we'll find many

benefits. Since there's such a huge number of IP addresses available, a

company won't really need to do NATing. Likewise, there can be more levels of

subnetting also possible.

Given all these benefits, there's some movement towards

implementing the protocol. The US federal agencies are supposed to shift to IPv6

by 2008. China, Korea and Japan have also been considering the shift to IPv6. In

fact, in July 2004, IPv6 records were visible for Japan and Korea's TLD (top

level domain) name servers became visible in the DNS root servers. The IPv6

records for France have also been added. Besides these, several key conferences

were hosted last year to create awareness about the protocol. Work is on at the

software front as well. Windows XP already has a trial deployment for it, while

some versions of Linux also support it. There's a company that's deploying a

DHCP server test suite for IPv6. The IPv6 forum has task forces from around 35

countries working on the protocol.

Despite all this action, IPv6 adoption is still a little

slow. Part of the reason is that it's not backward compatible with IPv4. An

IPv4 router won't support IPv6. It must have inherent support build in for

both protocols. Likewise, all applications and Operating Systems must add the

support for the new protocol. Just because this happens, it doesn't meant that

all companies and ISPs will immediately shift to the new protocol. It will take

time, as everybody would first have to understand the complexities and

challenges involved, do some pilot runs before doing the transition. Its

migration will be like any other technology. There will first be an intermediate

stage where, where both IPv4 and IPv6 will coexist. This will then slowly

transition to the new protocol. But the complete transition is not likely to

happen any time soon. Should you be worried about IPv6 now? Well, as there's

plenty of time for the transition to happen, you might like to consider doing

some dry runs of the same. It shouldn't happen that the world migrates to the

new protocol and enjoys its benefits, while you're left behind.

Anil Chopra, Associate Editor