 | Issue: | Asia-Pacific III 2013 |
| Article no.: | 8 | |
| Topic: | The world is running out of IPv4 addresses | |
| Author: | Raphael Ho | |
| Title: | Raphael Ho, Director of Network Engineering & Operations | |
| Organisation: | Equinix Inc. | |
| PDF size: | 206KB |
About author
Raphael Ho, Director of Network Engineering and Operations
Equinix Asia Pacific
As Director for Network Engineering and Operations for Equinix in Asia Pacific, Raphael Ho is responsible for the development of the Equinix Exchange platform and the Switched IP Connectivity Platform across the region.
Mr. Ho has over a decade of international management experience in the telecoms and enterprise business, with strong specialties in IP/MPLS and subsea networking, network operations and network management systems.
Aside from his work at Equinix, Mr. Ho is also the chair of the Internet eXchange (IX) Special Interest Group (SIG) at Asia-Pacific Network Information Centre (APNIC)*, where he facilitates the forum for sharing information about the status and activities of IX points in the Asia Pacific region to the peering community.
Prior to Equinix, Mr. Ho served as the Director for NOC Technical Support at Asia Netcom, supporting their transmission, IP and voice platforms. He has also served in various engineering/operations management positions in global organizations including Pihana Pacific, Level 3 Communications and Global One in Singapore, US and UK.
Mr. Ho holds a Bachelor of Engineering in Computing from Imperial College, London.
*APNIC is an open, membership-based, not-for-profit organization. It is one of five Regional Internet Registries (RIRs) charged with ensuring the fair distribution and responsible management of IP addresses and related resources. These resources are required for the stable and reliable operation of the global Internet. (website: http://www.apnic.net
Article abstract
In 1998, IPv6 standards were established, yet in 2013, IPv6 traffic accounts for only 0.1%. Large carriers have already deployed IPv6 in their 3-year cycle of core replacement, but consumer-facing ISPs see no consumer pressure. They have no business case if they have enough IPv4 addresses, and no incentive to upgrade, because IPv6 won’t inter-connect to IPv4 world. It will be the developing countries who will start mass migration, because their later broadband expansion involves installing latest IPv6-ready GEPON, and their stocks of IPv4 addresses are fast depleting. It is hoped that as more self-published content is native to IPv6 at the edge, consumers demand will grow, and the problems caused by carrier-grade NAT (CGN) will become apparent, thus encouraging all ISPs to upgrade.
Full Article
Fact: The world is running out of IPv4 addresses. Many industry experts are building models to predict when the last IP addresses will be consumed, and what is the best way to monetize the current scarcity of IPv4 addresses. There is no doubt that there is money to be made in this market.
Ironically, though, most of this money will likely be made from the world’s least developed countries, as they were not in the position to request large blocks of addresses while they were still plentiful. Today, these countries desperately need more addresses to cater to their population growth and support education and information and communications technology (ICT) initiatives that will enable their nations to compete in the new, hyper-digitized world.
The Internet Engineering Task Force (IETF), the architects of the Internet, realized this issue in 1992 and began development of the Next Generation Internet protocol, called IPng. Six years later in 1998 the standard was ratified as IPv6. The objective at the time was for everyone to migrate to a dual-stack network, in which every device would be able to speak both IPv4 and IPv6, knowing that IPv4 would depreciate at some point in the future.
That didn’t happen.
Fast forward to 2013. Today, IPv6 traffic accounts for approximately 0.1% of total Internet traffic . Network operators are increasingly looking at alternative technologies to extend the life of IPv4 (such as CGN, NAT-PT, etc), as well as the secondary markets for additional IPv4 addresses instead of deploying and upgrading their networks to support IPv6 natively.
Deploying large-scale network address translation (NAT) severely disrupts the any-to-any connectivity model of the Internet, which in the past has provided the fertile breeding ground for the Internet giants that stand today.
When you add this up, it equates to certain innovations being hampered, such as the Internet of Things, where every machine will be able to talk to any other machine in the world. It also means the barrier to entry in the global Internet marketplace will be disproportionately higher in developing countries, compared to the more developed countries. All of this is due to the unequal IP address distribution in the world.
Clearly, there is a technical need to migrate from IPv4 to IPv6 in order to maintain the any-to-any direct connectivity embedded into the ethos of the Internet engineers around the world. However, is there a business case to support this need? From a national or business viewpoint, if you are sitting on a large pool of IPv4 addresses, the IPv4 exhaustion doesn’t really impact you. On the other hand, if you have a deficit of IPv4 addresses, deploying an IPv6 only network does not solve the problem, as you’ll effectively be disconnected from the rest of the Internet running on IPv4. It will end up that you have to purchase more equipment to enable this connectivity. However, once this connectivity is built, the business case to deploy IPv6 to serve these customers is no longer valid, leading to a classic chicken and egg problem.
Is there a way out of this problem? Many governments have mandated that all of their own Internet properties and suppliers must deploy IPv6, however, many of these unfunded mandates have not been as successful as we have wished for .
Public awareness campaigns such as World IPv6 Launch was successful in gaining public mindshare and making news in almost every major mainstream news publication, but not in terms of actual traffic volume. However, any traffic growth is encouraging, and the group is optimistically predicting that IPv6 traffic will reach 10.8% by 2016 .
What’s most interesting is that most (if not all) international carriers have completed their IPv6 deployment at the core, but why did they do that? While “customer demand” is the most often cited reason, clearly there has not been any mass adoption, nor direct associated revenue of IPv6 from the customer base. Firstly, I believe this is because the carriers have a very aggressive technology upgrade cycle due to the explosive growth of traffic at the core. In fact, the entire core is essentially replaced every three years with the old core equipment pushed to the edge. The newer core equipment has hardware support for IPv6, and enabling IPv6 does not require any complicated business justification, making this mostly an engineering decision. Secondly, and perhaps more importantly, is peer pressure from other carriers, and technical one-upmanship within the Internet engineering community.
The situation is very different for consumer access, in which the deployed equipment often runs for decades. The ADSL access infrastructure in most countries has essentially been unchanged for decades, and the equipment does not support IPv6 natively. While there are migration tools and techniques for these legacy networks such as IPv6 Rapid Deployment (6rd), there are no demands from the end-users to force the ISPs to upgrade their infrastructure to run IPv6. Even when IPv4 addresses run out, it may still be cheaper for the ISPs to deploy large scale NAT to extend the life of the platform and to recover valuable IPv4 address space for resell or repurposing for the content servers.
From the content provider side, their interest is to deliver the content as fast and efficiently as possible to as many people in the world. Research has shown that even small latency and slower page loading times have measurable effects on revenue . Even today, the peering infrastructure for IPv6 is not as highly developed as the IPv4 infrastructure, leading to higher latency on average , as well as various other issues impacting end customer experience. For all these reasons, content providers have traditionally been reluctant to migrate their main web assets to IPv6 dual-stack, although this has changed after the last World IPv6 Launch day, in which many of the most visited websites committed to enabling IPv6 on their websites “for real” . However, this still only represents a small percentage of the content available on the Internet, and certainly not the killer reason for consumer ISPs to upgrade to IPv6.
So, what is the business justification for everyone to finally make the move to IPv6?
In my opinion, the developing countries will be the catalyst in the transition over the next few years. As they are late in the broadband deployment game, they have the opportunity to deploy newer technology, such as GEPON with the necessary IPv6 support at the head-end. Plus, coupled with the lack of IPv4 addresses, native IPv6 should be the logical choice for delivering services all the way to the endpoints. With sufficient users on IPv6, there will be more self-published content and applications that are exclusively on IPv6 in people’s dorms and garages, driving consumer demand. As the NATted IPv4 traffic grows, the performance limitation of the large scale NAT should convince more content providers to deliver content natively on IPv6 by dual-stack enabling their servers.
With the National Broadband Network’s (NBN) plans, proposed by various governments, to replace the aging copper infrastructure and upgrade to newer equipment that supports IPv6 at no extra cost, this virtuous cycle may finally break the chicken and egg problem.
With the additional address space provided by IPv6, each ISP subscriber will have enough IP addresses to address 80 Octillion devices in their home with the ability for each one to talk directly to the potential 340 Undecillion devices on the Internet, which certainly has potential that cannot be imagined today.
I look forward to the day when we finally complete the migration and switch off IPv4 on our routers.
