IPv4 or IPv6—Myths and Realities

This chapter is from the book 

The year is 1977. Earth's population has not yet reached 4.5 billion. One hundred and eleven interconnected computing machines make up the ARPANET, a research network.

Thirty years later, in 2008, Earth's population peaks at 6.6 billion and the Internet, with a population of 1.3 billion, has yet to reach 22 percent penetration rate, the threshold that qualifies it as a massively adopted technology. While arguing about the lifetime scope of the available IPv4 address space, the Internet community aggressively pursues a massive convergence of communication technologies (audio, data, video, and voice) over IP. The community is still debating the urgency of an upgrade to IPv6.

In the year 2030, Earth's population is expected to be over 8 billion, adding nearly 75 million people every year, or twice the population of the state of California. The Internet is an integral part of the worldwide economy and everybody's life. The old IPv4 versus IPv6 debate is now history.

The Business Case for IPv6

To a large degree, mass adoption of new technology is fueled by a person's vision of "What's in it for me?" Can the new technology improve my business operations? Can I use it to provide a new profitable service? Is adoption needed to stay competitive? Will the new technology enrich my personal life?

At the end of the '70s, few of the IP designers envisioned the rapid and widespread adoption of IP; IP became the convergence layer for communication services in many industry segments such as home, mobile wireless, transportation, media, and many others. This convergence, along with a plethora of new Internet-enabled devices, provides a fertile and unexpected foundation for innovation that far exceeds the original design constructs. Information movement is now the game, and content is king.

So is an Internet upgrade necessary to sustain the growth of the future and to interconnect all the devices of the new global economy? Will IPv6 provide the fire to fuel the growth?

Before debating the pros and cons of the new IP version, let's look at the historical perspective of IPv6 and its development.

A Brief History of IPv6 Standardization

At the end of the '80s, the Internet Engineering Task Force (IETF) began to evaluate the consequences of the Internet's growth on the protocol, with particular emphasis on addressing. The organization evaluated:

• Address space exhaustion: The original IPv4 addressing plan was mathematically limited to 65,536 Class B networks for the entire Internet. The assignment rate of the former Class B networks (blocks of 65,536 contiguous addresses) would lead to the exhaustion of IPv4 addresses sometime close to 1994.
• Expanding routing tables: The allocation of Class C (blocks of 256 contiguous IPv4 addresses) networks instead of Class B networks would lead to an alarming expansion of the routing tables in the Internet backbone routers—typically Cisco AGS+ or 7000 series.

In November 1991, the IETF formed the Routing and Addressing (ROAD) working group (WG) to analyze and deliver guidelines to address these issues. In March 1992, the WG provided its recommendations in two categories:

• Immediate: Adopt the Classless Interdomain Routing (CIDR) route aggregation to control the growth rate of routing tables and allow finer-grained allocations than previous 8-bit boundaries defined as Class A, B, and C.

  IPv4	CIDR	IP
  Class	Notation	Addresses
  A	256	16,777,216
  B	65,536	65,536
  C	16,777,216	256

• Long term: Initiate a call for proposals "to form working groups to explore separate approaches for bigger Internet addresses."

At the beginning of the '90s, the use of the Open Systems Interconnection (OSI) reference model's network and transport layers was heavily promoted through the U.S. and UK Government Open Systems Interconnect Profile (GOSIP). In the end, it failed to get widely deployed due to the lack of applications running over OSI. Nevertheless, by mid-1992, the Internet Advisory Board (IAB) proposed, as an immediate solution, the use of Connectionless Network Protocol (CLNP), which would be the basis for a next generation IP, naming it IP version 7. This proposal was highly debated because OSI was not viewed favorably at the IETF. The IAB recommendation was rejected by the IETF, which called for a number of working groups to work on candidate proposals. In 1993, an IETF IP Next Generation Decision Process (ipdecide) Birds of a Feather (BoF) session set the criteria that would drive the definition of the new protocol. The end result was the creation of an Internet Protocol Next Generation (IPng) directorate that was tasked to

• Define the scope of the IPng effort, keeping in mind the time constraints
• Develop a clear and concise set of technical requirements and operational criteria for IPng
• Recommend which of the current IPng protocol candidates to accept, if any

Four parallel projects began exploring ways to address the identified consequences of the rapidly growing Internet:

• CNAT: Tivoli's Comprehensive Network Address Translator.
• IP Encaps: The proposal evolved to become IP Address Encapsulation (IPAE) and then merged with the SIP proposal.
• Nimrod: A proposal viewed as a research project by the Internet Engineering Steering Group (IESG).
• Simple CLNP: The proposal later became TCP and UDP with Bigger Addresses (TUBA).

Three additional proposals were later brought into the discussion:

• The P Internet Protocol (PIP): The proposal merged later with SIP and the resulting working group called itself Simple Internet Protocol Plus (SIPP).
• Simple Internet Protocol (SIP): The proposal evolved to become IP Address Encapsulation (IPAE) and later merged with the SIP proposal.
• TP/IX: The proposal was later renamed Common Architecture for the Internet (CATNIP).

All the work that went into these projects and the resulting mergers was finally evaluated by the IPng. Three proposals were retained: CATNIP, SIPP, and TUBA. As documented in RFC 1752:

• None of these proposals were wrong nor were others right. All of the proposals would work in some ways providing a path to overcome the obstacles we face as the Internet expands. The task of the IPng Area was to ensure that the IETF understand the offered proposals, learn from the proposals and provide a recommendation on what path best resolves the basic issues while providing the best foundation upon which to build for the future.

After countless discussions and reviews of the strengths and weaknesses of updated versions of the submitted proposal, the consensus of the IPng Directorate was to recommend that the protocol described in the SIPP specification, which began as 64 bits and evolved to 128 bits, addressing should be adopted as the basis for IPng, that it should be the next generation of IP, and that is should be named IP version 6. The recommendation for IPng was approved by the IESG and became a proposed standard on November 17, 1994, as RFC 1752. This new version of IP can be considered an evolutionary step rather than a revolutionary step in the development of IP. Some of the principles that guided the changes are to

• Keep all aspects and features of IPv4 that were proven to work and continued to make sense
• Remove or make optional all features of IPv4 that were infrequently used or shown to be problematic
• Add new solutions to fix existent problems or add new features that enable the protocol to address new needs

The core set of IPv6 protocols was made an IETF Draft Standard on August 10, 1998, an event that represented the green light for vendors to develop their implementations and submit their code for interoperability testing. From 1996 to 2006, the experimental 6bone (http://go6.net/ipv6-6bone/) overlay IPv6 infrastructure offered the infrastructure framework for wide interoperability tests. In 2001, IPv6 started to be integrated on commercial products such as Sun Solaris 8, Cisco IOS Release 12.2(2)T, and Juniper JUNOS 5.1. The indication that IPv6 is technologically ready was the IETF intent to close or recharter the IPv6 WG in December 2006.

Is IPv6 ready for deployment in your business? Why should the world care about IPv6 today?

Looking at the Numbers

Initially, one of the main objectives of the IPng effort was to identify ways to cope with the explosive growth of the Internet. Today, this growth continues at a faster rate, reaffirming the premise of the IPng work. Making a business case for the new protocol comes down to a review of the numbers. From a global perspective, these numbers were already described by one of the authors in the "e-Nations, The Internet for All" paper, which was endorsed by the United Nations.³

The Internet—an ever growing and widely popular environment for communication, information sharing, and collaboration—could simply not be promoted as a mass-market technology. In addition, the foundation of the worldwide economy would not work if the Internet's base protocol (IP) did not offer the necessary address space resources to equitably connect the population of every country around the world.

The expansion of the Internet is also tied to the rapid development and market penetration of enabling technologies such as high-speed broadband and wireless access. Many enterprises have shifted from point-to-point, ATM, and Frame Relay infrastructures to IP-based local- and wide-area networks (LAN and WAN) for basic business operations. Traditional voice carriers are migrating their voice network to IP-based transport to reduce or eliminate future capital expenditure (CAPEX) and operational expenditure (OPEX) related to redundant parallel network infrastructures. These IP-based technologies modify an application's landscape by changing the use of the Internet from a client/server model to a more distributed model or peer-to-peer model. Very rapid and successful adoption of distributed applications such as Voice over Internet (VoIP), instant messaging, content sharing, and Internet gaming leads people with "always-on" and "always-best" access to the Internet to be content producers as well as consumers. An expanded IP address space is necessary to support this paradigm change in the way the Internet is used.

Lack of IP resources can lead to an increasing digital divide between information and communications technology (ICT) rich and ICT poor countries. So let's have a look at those "numbers" that make IPv6 a "must."

Earth Population Versus Internet Users

By the end of 2007, world population reached over 6.6 billion humans⁴ and a United Nations report forecasts an increase to over 8 billion by 2030. Although the Internet is deeply embedded in the worldwide economy, it reaches only one-sixth of today's population with 1.3 billion users, as shown in Figure 2-1.

Internet usage has seen accelerated growth across the world, particularly in emerging markets. For example, Africa, the region with the least Internet penetration, has seen the usage grow over 880 percent between 2000 and 2007. To provide equal opportunities worldwide, the Internet architecture must cope with rapid growth in consumer interest and usage. The forecast for growth leads to a new perspective on the demand for IP address space. Even without taking into consideration expected address allocation inefficiencies, IPv4's 32-bit address space is inadequate to support a plethora of connected devices owned by one-third of Earth's population.

When accounting for expected growth, 50 percent of the worldwide population ends up without IPv4 address space to connect appliances to the Internet. Table 2-1 provides an analysis of the address space necessary to achieve 20 percent Internet penetration in each world region (expected growth has been accounted for).

Table 2-1. The Population of World Regions and the IP Address Space Needed to Cover 20 Percent of the Population

Region	Population	Number of /8 Subnets Needed for 20% of the Population with 1 Address per Person (HD Ratio 90%)
Africa	941,249,130	93
Asia	3,733,783,474	431
Europe	801,821,187	78
Latin America/Caribbean	569,133,474	53
Middle East	192,755,045	16
North America	334,659,631	30
Oceania/Australia	33,569,718	2
World	6,606,971,659	808

As Table 2-1 indicates, as of February 2008, the world requires 808 IPv4 /8 subnets, more than twice the possible 256 /8 subnets, for the Internet to be considered a massively adopted technology. The IPv4 address space clearly cannot sustain the Internet's penetration worldwide.

The analysis in Table 2-1 assumes that each Internet user owns a public address. While this becomes a necessity for the latest usage patterns and the new peer-to-peer applications, it was quite common to have multiple Internet users sharing a global IPv4 address when dial-up was the main technology to connect to the Internet.

Highlighting the developing digital divide, it should be noted that as of June 2007, the population of the top 22 countries in Internet penetration represents 10 percent of the world's population.⁶ The Internet reached mass-adoption levels in only 99 (40 percent) of the world's 245 countries.

Mobile Phone Market Segment

For the past 15 years, Global System for Mobile (GSM) communications, along with other cellular technologies, has dramatically transformed daily life for billions of people. From Q2 CY07, the number of GSM connections, as shown in Table 2-2, has grown to pass the 3 billion mark in April 2008 globally, as announced by the GSMA, the global trade group for the mobile industry⁷

Table 2-2. Number of GSM Connections Per Region^a

Market	Connections in Q2 2007
World	2,377,790,703 (out of 2,831,345,390 wireless subscribers)
Africa	220,734,625
Americas	252,371,017
Asia Pacific	917,356,568
Europe Eastern	359,637,084
Europe Western	387,248,744
Middle East	146,458,459
USA/Canada	93,984,206

Internet applications and services are not only possible via the public Wi-Fi and upcoming WiMAX infrastructures; they are also fully integrated, including IPv6 support, in the third and fourth generation telephony through the IP Multimedia Subsystem (IMS). The new generation of wireless devices comes with an embedded dual IP stack and multimedia applications, including VoIP. The fierce competition between content providers seeking new revenues and increased market shares is leading to the delivery of new content and services over IP that will rely on always-on connectivity and end-to-end reachability. The combination of wireless and new broadband technologies such as DOCSIS 3.0 for cable or fiber to the home (FTTH) is leading to more and more independence of the service offering from the type or point of access and drives the market toward the convergence of fixed-mobile services.

If just 50 percent of worldwide subscribers transition to those new technologies and services, they will require an additional 66 /8 networks for always-on connectivity. This example does not take into account the forecasted increase in the number of subscribers and the addresses required by the infrastructure supporting all these users.

Consumer Devices

The digital revolution that marked the end of the previous millennium brought a wide variety of devices into our lives. Although they entered the market as "gadgets," many of these devices quickly became indispensable to many people. Gaming consoles (more than 150 million, including more than 44 million Sony PS3 and PSP), multimedia players, digital video recorders, digital cameras, and Global Positioning System (GPS) consoles are just a few examples of the many devices that are no longer a novelty.

The power of these new devices does not reside in their standalone operation but rather in the services they can offer when connected to other devices. The integration of IP over Ethernet and wireless technologies provides an environment where consumer devices can easily access resources and services. In order to communicate, these connected devices each use at least an IP address. Moreover, for full service and business model flexibility, these devices require public IP addresses. Their rapid adoption represents yet another source of pressure on the IPv4 address space.

Connected homes and public wireless LAN services represent perfect infrastructures to proliferate IP-enabled consumer devices. Although it is difficult to track such a diverse set of products, it is estimated that in 2006 there were 492 million connected consumer devices such as phones, computers, game consoles, and media centers. By 2010 that number is expected to reach 2.8 billion units.⁸ At one address per device and an HD ratio of 90 percent, these connected devices require 271 /8 prefixes (surpassing the total IPv4 address pool) and would need 1871 /8 prefixes by 2010. Many of these consumer devices could reuse private IPv4 addresses but this would limit the type of services available and the flexibility to adopt new business models while also increasing the cost of the applications supported.

The number of consumer devices, their need for global reachability, and their expected mobility outside of the home require a significantly larger address space than what IPv4 can offer. Unfettered growth and large-scale adoption are essential in this market space as it stimulates new service concepts and product innovation based on consumer requests. IPv6, with its large address space, is the natural answer to this market's IP address needs. At the same time, IPv6 offers specific features, such as stateless autoconfiguration, that can reduce product costs, a great asset in a low-margin market.

Transportation

A significant part of our day depends to a certain extent on one form of transportation or another. Public or private transportation takes us to and from our place of work; transportation provides the logistics that support our global economy; or perhaps transportation is the very scope of our business. Transportation can also make vacations possible or frustrating. In summary, we depend on various forms of transportation in our daily lives and the means by which we travel have us as a captive audience for a significant part of our day. The combination of wireless access and IP connectivity can provide significant business and increased revenue opportunities in the transportation market. Following are some opportunities for revenue:

• Telematics: Sensors distributed in a vehicle can monitor and manage its operation, providing new services to the vehicle owner, including the data for improved maintenance and troubleshooting. In late 2007, BMW's Research and Technology division unveiled its iDrive pilot program, which integrates the large number of control systems and entertainment systems through an integrated IP-based network. BMW's goal is to use a standards-based platform for future anticipated needs, simplify development and manufacturing, and reduce long-term costs. Rail systems are using telematics to manage spacing between trains to maximize passenger loads and improve safety.
• Vehicle to vehicle: Along with the development of telematic applications, communications between vehicles could be developed in conjunction with road infrastructures that work together to improve safety and prevent accidents. This type of environment integrates a wide range of wireless/wireline communications and control technologies in a framework developed by the Intelligent Transport Systems (ITS) standards (ISO TC 204).
• Fleet connectivity: Transportation companies can leverage municipal Wi-Fi LANs and cellular broadband to connect their assets back to the central office. It is an effective and cost-saving mechanism to coordinate activities, synchronize inventory, and update routes. E-ticketing, real-time information for passengers, and video surveillance are typical applications that benefit from the availability of Internet access on public transportation. The cost of deployment can be covered by additional services such as local advertisements and news contracts negotiated with appropriate channels.
• Internet access "on the road or in flight": Inside their own cars, on public transportation, in airplanes, or aboard cruise ships, people represent a trapped audience that will pay a premium for access to content whether it is for work or entertainment.
• First responders fleet: This is another market segment that could benefit from bidirectional communications for applications such as video and database access. There is great interest in the integration of all assets that need to be leveraged in case of emergency. Recent press highlighted innovative communities deploying metro wireless infrastructures that could be used by the emergency responders. These new infrastructures lead to radio frequencies traditionally used for those communications to be freed up for other usage. Two notable initiatives are working on the future communications infrastructures for first responders: U-2010 (http://www.u2010.eu/) and MetroNet6 (http://www.metronet6.org/).
• Cargo monitoring: Tracking goods in transit is becoming more and more important to provide proper environmental conditions (maintaining temperature levels for perishable foods) and to constantly monitor valuable goods.

Cars, ships, trains, and airplanes have long-lasting power sources and have no major constraints related to the size of the communications devices they can be fitted with. This makes them ideal environments for mobile communications services. It is expected that vehicles will support multiple IP-connected devices, so they will require entire IP subnets to support them. They must also be able to connect seamlessly to various access network types such as wireless services. It should not be expected that a single access media type or access provider can cover all countries or regions or cities. The need for this type of flexibility also makes the case for the use of IP mobility.

It is rather difficult to evaluate the volume of addresses that would be used by networked vehicles but a recent study about the European market forecasts the numbers to be in the millions range. Table 2-3 provides a summary profile of the European road-based transportation.

Table 2-3. European Market Size for Road Transportation^a

Vehicle Category	Vehicle Type	Number of Vehicles	New Vehicles per Year	Vehicle Lifetime (Years)
Public
Pro Vehicle	Police	200,000	40,000	5
Pro Vehicle	Ambulance Taxi	15,000	3,000	5
High End Vehicle	Bus	175,000	35,000	5
High End Vehicle	Fire (>16t)	32,000	7,000	5
High End Vehicle	Full Ambulance	20,000	4,000	5
Large Vehicle	Metro	20,000	700	30
Large Vehicle	Reg&Sub Rail	55,000	2,000	30
Large Vehicle	Light Rail	25,000	1,000	30
Private
Pro Vehicle	Car	220,000,000	17,000,000	10+
Pro Vehicle	Goods Vehicles	20,000,000	4,000,000	5

The 2006 data presented in Table 2-3 indicates that if an IPv4 /24 subnet is used per vehicle to interconnect its various sensors and communications devices, a deployment target of 5 percent of the European transportation market alone will require 183 /8 subnets.

The transportation market space is full of opportunities for new communications services. Cruise ships are fully networked and use services such as VoIP internally. Airplanes provide Internet access services, and multiple automakers are piloting networked cars. Table 2-3 indicates that the life cycle of a vehicle is generally long, between 5 and 30 years. Older OEM vehicles may never be updated. Others will be retrofitted with newer in-transit systems where there is business value such as safety, security, or attracting customers.

Industrial Sensors and Control Systems

Industrial networks (building, plant, and process automation networks) are migrating from legacy techniques to reliance on IP-based services, as shown in Figure 2-2. The drivers for change are economics, interoperability, simplification, and common cross-network security enforcement.

The more sensors that are used in the manufacturing process and in tracking a product's path through the distribution chain, the more optimizations can be identified and applied to each step of the process, as shown by the European Reconfigurable Ubiquitous Networked Embedded Systems (RUNES) project (http://www.ist-runes.org). Interconnecting sensors into a consolidated product management framework leads to significant productivity increases and cost reductions. They can also enhance security and management of fixed assets. Sensors can be deployed internally by enterprises, but we expect their footprint to grow with more and more sensors deployed in public domains, modes of transportation, and homes.

The migration of industrial sensors and control systems to an IP-based architecture is once again the result of several technologies:

• Back-end and front-end control systems: Applications running on computers and exchanging data through an IP network
• Industrial sensors: Span a wide range, from passive radio-frequency identification (RFID) with no IP address to Motes (small wireless transceiver attached to a sensor) or smart cards with an embedded IP stack
• Readers or gateways: Devices that collect data from sensors over specific wireless technologies; for example: IEEE 802.15.4 (low-rate wireless personal area network) with an embedded IP stack

To help the creation of an open and standardized architecture for sensor-enabled systems, the IETF IPv6 over Low power WPAN (6LOWPAN) working group⁹ leveraged IPv6 to solve challenges such as self-configuring networks, an aspect very typical to sensors' environments. Management and access of industrial sensors will be done both within the LAN and over the public domain, driving the need for IPv6 capabilities such as address space, "plug-and-play" autoconfiguration, communities of Interest, and so forth.

As shown in Figure 2-3, an estimated 127 million wireless sensors are expected to be deployed by 2010.¹⁰

At least 12 /8 prefixes are required to connect these devices. Wireless access facilitates the deployment of sensors and thus helps accelerate their adoption, which in turn increases the demand for IP addresses. IPv6 is perfectly suited for this market space. It has the necessary address space to cover a large number of devices and has the tools necessary to provide for simple provisioning of this type of devices, which generally have little processing power.

Common Observations When Looking at the Numbers

Interestingly, as soon as the depletion process of the IPv4 address space was slowed down through various conservation and management mechanisms, the immediate interest in its successor diminished. For the years that followed, the search for a reason to invest in an IP upgrade to IPv6 focused mainly on the application layer. The thorough scrubbing of IPv6-specific features and the brainstorms of IPv6 enthusiasts have yet to produce a killer application that would trigger market adoption. But, did we really make the most out of the last killer app we came up with, the Internet? The true potential of the Internet and of IP has yet to be unleashed, and this cannot happen in the context of its initial definition.

This chapter intends to show the technical arguments related to the new protocol. By looking at just a few statistics, we highlight the basic resource requirements for the continued growth of current markets. Some of the estimates presented here are backed by formal reports of address shortages. For example, the large cable providers in the United States reported running out of private IPv4 addresses in 2005.

Innovative applications that people will later call "killer apps" will certainly come with the IPv6 protocol. For now, however, just the basic market needs make a strong case for IPv6, which provides:

• Resources to scale up current networks: The larger address space is mandatory to meet current numbers of devices and to support the expected Internet population growth.
• Resources to simplify network and service architecture: Network and service design constraints due to address shortage can be eliminated, leading to reduced costs of operation.
• An environment for continued innovation: A larger and simpler Internet that integrates ever more diverse devices represents an environment that stimulates innovation, which in turn stimulates adoption.