The technology on which the internet is built, IPv4, is cracking
under the strain of exploding demand for web utility. Now IPv6 is
here, but will it ease the burden?
Although statistics can always be spun to suit the latest marketing
opportunity and the IT business falls second only to politics in
this respect, it is hard to ignore the facts when it comes to IPv6.
The Internet is built on IPv4 technology and at 30 years old it is
not surprising that it is creaking under the strain of dramatic
surges in demand for access, and things will only get worse as 3G
applications and services start to eat into the addressing sector.
With 70 per cent of all Internet addresses utilised by the US,
leaving the rest of the connected world to share a meagre 30 per
cent, it is not surprising that technology leaders such as Japan
are not only taking an interest in IPv6, but actively making it a
reality in the here and now. Nearer to home, the Euro6IX project -
next generation Internet services based on IPv6 - is being backed
by the European Commission.
Of course, the idea that IP address space is quickly running out is
nothing new. According to differing sources, it has been rapidly
depleting for years. So, with the Internet imploding, is IPv6 its
saviour?
"IPv6 provides a solution to the address crunch by increasing the
address space from 32 bits to 128 bits. This value is estimated to
accommodate ten billion devices over the Internet and although this
may seem far-fetched, the reality is that computers are
incarcerated into almost every facet of our lives, including
household devices, PDAs and electronics," says Uri Rahamim, vice
president of Hitachi Internetworking.
Naturally, all of these connected gadgets will need an IP address,
and then there's the small matter of mobile telephone use.
Mobiles are becoming smarter by the day and Internet connectivity
is a driving force. Throw 3G always-on access and the rapidly
growing number of mobile users year-on-year into the mix and it
doesn't take a genius to realise the problem of limited address
space is very real indeed.
Dire situation or red herring?
Rahamim reckons "the
current IPv4 scheme of four billion addresses will soon be
exhausted". But it is a view not necessarily shared by all. Array
Networks director of product management Steve Shah claims "when
alternatives to IPv4 were initially being sought, most folks were
giving every machine a real, routable IP address. At the rate that
was happening, there was a genuine problem. However, three
technologies changed that: HTTP/1.1 (HyperText Transfer Protocol),
NAT (Network Address Translation) and CIDR (Classless Inter-Domain
Routing)".
HTTP/1.1 found a solution by providing virtual hosting, so single
servers were able to host thousands of Web sites on a single IP.
"Countless hosting providers are 'garage ISPs' that use a Linux
server or two to dish up content from thousands of virtual hosts
off one or two IP addresses. Why use more when you don't have to?"
Shah argues.
Why indeed, which is where NAT comes in to change the dynamic for
many businesses, making it possible for organisations with tens of
thousands of staff to comfortably fit behind a small handful of IP
addresses. Of course, saving IPs was the side effect of people
needing firewalls to segregate networks from the 'harsh' Internet.
"Finally, CIDR made a significant impact on IP network allocation
since it became possible to segment networks into subnets with as
little as two IP addresses," Shah continues. Hosting providers were
finally able to truly make use of a single class C network without
wasting hundreds of IPs when customers wanted their own private
subnet. CIDR also made a difference for those providers with class
A and B networks. "Now those networks could be effectively split up
and given to individual customers without relinquishing more than a
few IPs," Shah concludes.
Phil Goff, technical director at Allasso, agrees the address space
issue has diminished somewhat in recent years because a lot of
organisations are using private addressing schemes and firewalls to
translate into real addresses. "Practically, this means that
although they may have hundreds or thousands of IP-based systems
internally, they need very few external or public addresses. The
upshot is that the rate at which addresses are allocated by the
Internet authorities and ISPs has reduced significantly," he
argues.
v4 versus v6
So if the rule of diminishing IP addresses
is something of a red herring now the dynamic of Internet use is
changing, where else does IPv6 have a role to play? Alexander
Helmke, senior product manager at Wind River Networks Business
Unit, lists some of the shortcomings he sees inherent in
IPv4:
- Its limited address space hinders further growth of devices
attached to the Internet
- It has insufficient end-to-end security, making the Internet
less of a trusted medium
- It requires complex procedures to administer and manage the
network
- It is unable to support varying qualities of service and
different media types such as data, voice and video
As Rahamim says, these inadequacies are a legacy of IPv4 being an
'old' protocol conceived more than two decades ago. "IPv6 fixes
problems such as class-based/classless addressing and growing
routing table sizes in the Internet backbone. Its hierarchical
addressing scheme and improved header will allow for more efficient
IP packet processing," he insists.
Of course, IPv6 also brings many improvements in areas such as
routing, simplified network auto-configuration, integrated quality
of service for multimedia and voice applications, and heightened
security. Yet critics such as Shah remain defiantly sceptical.
"The problem with IPv6 is that it is complicated, significantly
more so than IPv4. Look back at any wildly successful technology.
Ethernet is an excellent example. There were at one point countless
competitors for the technology that were truly better. What made
Ethernet succeed was its simplicity in implementation and
deployment. IPv4 is another example of this - it is easy to
configure, relatively easy to build and very widespread," he says.
Testing the water
In light of all this debate, which pretty much reflects the
uncertainty and confusion regarding IPv6 at grass roots level, one
might be excused for thinking it will remain a well talked about
technology but never become a real world one. Excused, but wrong.
Initial rollout in the UK has already started.
"The first commercial IPv6 service network, Telia/Skanova, was put
in place last year using Hitachi Dual-Stack (IPv6/IPv4)
gigabit routers. These routers can handle IPv4 and IPv6 traffic at
full speed," reveals Rahamim. Other carriers including British
Telecom have gone through evaluations and initial trials with IPv6
products and technology - some are even providing small scale IPv6
services. It is anticipated that several carriers will have
completed initial IPv6 field trials by early to mid-2003 and be
ready for full deployment soon after.
Overseas, particularly in the Asian market, things are also
starting to get done on the ground. Michael M. Holliday, senior
product manager for IP Transit Service at Global Crossing, predicts
the first mainstream commercial IPv6 services will likely be in
Japan, which he sees as very aggressive in its implementation of
the new protocol.
"This is due in part to mandates by the Japanese and South Korean
governments regarding adoption of IPv6 by government institutions,
educational facilities and large industrial concerns by 2005.
Europe follows Asia, but in both cases one of the driving forces is
the launch of 3G wireless. Considering that IPv6 is an important
element in the operation of 3G devices, regarding the additional
addresses required and the mobile routing aspects, deployment of it
will need to proceed apace with the commercial rollout of 3G
services," he explains.
Only fools rush in
One thing you can be crystal clear
about is that transition to IPv6 will not happen overnight - far
from it. The process will take several years depending on the scale
of deployment. But key to successful transition is the maintenance
of full backwards compatibility with IPv4.
For example, Hitachi's GR2000 has support tunnelling mechanisms
that use existing IPv4 routing infrastructures to carry IPv6
traffic, and its routers implement a dual-stack approach, allowing
old and new networks to run simultaneously within the same box.
Although the technological waters are far from clear where
upgrading to IPv6 is concerned, only a fool would suggest sitting
on one's hands in the dark and hoping the very real Internet
addressing problem will simply go away. While we shouldn't rush
into an IPv6 upgrade as a matter of urgency just yet, there is
certainly plenty of merit in at least exploring the options and
studying its business benefits and impact on budgets.
We all know the IT industry hums along to the mantra of 'if it
ain't broke, don't fix it', but many of us seem content to ignore
the fact that IPv4 is most definitely damaged beyond repair. Early
adopters may not see a black and white ROI for their efforts, but
reaping the benefits of competitive advantage and avoiding the
costly last minute rush towards a solution that befell so many
slack players approaching Y2K is not to be sniffed at.
A vendor's eye view of IPv6
"Carriers and service
providers need to start developing an upgrade strategy to migrate
to IPv6 now. Upgrading is not a trivial task and users should start
to incorporate and test IPv6 into their existing network
infrastructure. It is better to start sooner than later," says
Hitachi Internetworking vice president Uri Rahamim.
"Without disruptions to your business, a dual-stack router which
can support IPv4 and IPv6 simultaneously will allow you to continue
to run your day-to-day IT operations while slowly migrating users
to IPv6. This should be done in phases and the use of tunnelling
techniques will make this transition transparent to users.
"The deployment strategy will more than likely dictate the costs
associated with rolling out IPv6. But given the availability of
dual-stack routers and tunnelling mechanisms, the rollout should be
transparent from the user point of view. Users can continue to use
their existing router even after the transition to IPv6.
"To avoid being unprepared when IPv6 capabilities come into demand,
carriers and ISPs need to upgrade their equipment now, or face
being left with useless IPv4 equipment in 18 months time. The costs
and technical issues are minimal compared to the ROI.
"In the long run, IPv6 reduces network administration with its
support for auto-configuration. This simplifies the tedious task of
having to configure IP addresses manually during upgrades, moves
and network expansion. The increase in the availability of IP
addresses alone will enable new applications for consumers.
"Increase in IPv6 performance can be used to deploy multimedia
business applications, thereby expanding corporate revenue
streams."
www.allasso.comwww.arraynetworks.netwww.globalcrossing.comwww.internetworking.hitachi.com/www.windriver.com/