Sometime this year the world will run out of Internet addresses doled out under the old-style IPv4 protocol, and while the 128 bit addresses in IPv6 will allow for essentially an unlimited number of addresses, the upgrade path is a tricky one.
For starters, the protocols are incompatible in a binary fashion. Network equipment designed to run IPv4 cannot effectively route IPv6 Internet traffic. IPv6 gear typically will be backward compatible and will be able to handle IPv4 packets; however translating packets from one world to the other is still not trivial.
That adds up to a backwards compatibility problem of epic proportions. Think of the impact of this in terms of broader network interoperability. The Internet community and its various providers of connectivity, transit, collocation, and other services, are far from monolithic. The whole Internet won't flip the switch one day and transfer everything to the IPv6 protocol. Guaranteed, in that awkward period of transition, connections will break. We've never made this big a switch.
Beyond compatibility, IPv6 will also give large swathes of the switching and routing infrastructure of the Internet a crushing performance downgrade. Many of the core, critical packet traffic cops use customized ASICs designed to accommodate IPv4 data streams. The manufacturers of these devices have given them the capability to support IPv6 by means of software upgrades. But purpose-built ASICs cannot be upgraded, so the tasks of routing will fall on the CPUs instead.
Further, the typical routing device does not tend to have a particularly powerful CPU. After all, that's what the ASIC is for, right? The result? Routing IPv6 traffic will dramatically reduce throughput speeds. Routers and cards that once ran on cruise control for days will begin to stagger under the increased performance requirements of IPv6. Your mileage may vary, but experts estimate performance declines of as much as 60% may loom.
Then there are the issues of end-user experience and Web site performance. A significant portion of the wireless routers and home networking equipment is not designed to handle IPv6. While modern operating systems have IPv6 compatibility, millions of users are on older OS platforms that will require continued IPv4 support. As a result, publishers of content, providers of connectivity and advertising networks will all need to accommodate this dual-mode reality for years to come.
Search engine rankings could be scrambled, as well. Google is aggressively moving towards IPv6 and its possible that future search result pages will show a strong bias towards IPv6-enabled web servers and Web sites. Already when accessing Google through IPv6, the search engine will show results for sites supporting IPv6. For those sites that do not support the new upgrade, IPv6 renders it effectively invisible for the IPv6 users (even if the user can access it through some IPv4 gateways).
Add this all up and there are multiple implications for IPv6. First, network operators from the biggest to the smallest will need to closely coordinate their efforts to ensure that the Internet maintains wide-scale connectivity and IP compatibility. These same network operators must embark on an aggressive network upgrade path. Finally, network operators, publishers, e-commerce purveyors, advertisers, ISPs must plan for a double-existence and an architecture that supports both IPv4 and IPv6 for the foreseeable future.
Sounds expensive? It can be. One way that many ISPs are planning on handling the problem is by "dual-stacking" - maintaining network equipment that runs in parallel with one handling IPv4 traffic and the other handling IPv6 traffic.
Maintaining an entire network with Internet content along with a data delivery infrastructure while having one foot in both worlds, is both cumbersome and complicated, not to mention costly. Putting in place transparent proxy servers that can translate traffic into the appropriate IP version in order to meet user needs could help alleviate the issue. However, end users trying to reach a web site could still fail to connect if the Domain Name Servers of the destination site or network are using the wrong version of IP, causing the connection to break before the user even reaches the web server.
An alternative solution is to offload most of the IPv6 burden to a Content Delivery Network (CDN). The CDN would be tasked with delivering the entire Web site for a customer to all externally generated page requests. In this scenario, the CDN will serve as the public IPv6 face of an advertising network, e-commerce site, content publisher or social network. The CDN will accomplish this translation and presentation at the content level rather than the packet level.
As a result, objects and other key data elements will be retrieved from origin servers via IPv4 but will be served to the end user via IPv6. This system will minimize confusion and compatibility problems not only with search engines and ISPs but also with DNS servers, which will see the CDN as presenting a valid IPv6 address and treat it accordingly.
Equally important, the same CDN could be used in the opposite fashion, to present IPv4 traffic to end users or networks still operating on the older protocol. By using a CDN as a buffer, companies that want to both serve early adopters on IPv6 and keep the very late adopters on IPv4 happy will be able to keep everyone happy.
Eventually, the need for this uncomfortable dual existence will go away. But it will likely take several years for the IPv6 transition to permeate the entire Internet infrastructure and user base. Until then, the Web will be a bipolar world of old and new living side by side and in the fast and slow lane, simultaneously.
The best initiative for IT teams to employ is to examine all the possible options for bridging IPv4 and IPv6 in the interim with an eye to the future for transitioning the whole network over to IPv6. Commit your infrastructure vendors to make that transition with you each step of the way, with no interruption to the business. Although a challenge, this metamorphosis will result in better security, faster transactions and will guarantee that customers will connect with your web site.
Safruti is vice president of product strategy at Cotendo and Kuperman is vice president of operations at Cotendo, an innovative provider of Content Delivery Network and Site Acceleration Services located in Sunnyvale, Calif.