IPv6: Coming Soon to a Network Near You
The long-term benefits far outweigh the immediate difficulties.
It’s no secret that in the coming weeks, IP version 4 will be exhausted. IPv6 will provide the address bonanza that will simplify the transition to IP for cable system operators in the coming years. What’s been less apparent is the additional feature portfolio that IPv6 will bring to the industry.
For operators that move quickly to begin the transition and that utilize the full value of the new protocol, IPv6 can be highly rewarding for many years to come.
As one who has been conducting training sessions to prepare cable for the arrival of IPv6, I’ve seen the imminent arrival of IPv6 anticipated in the same way that most of us view road construction: On one hand, we know the process is going to have more than its share of difficulty; on the other, we will be grateful for the long-term improvement.
Still, implementing a new address plan for the cable network is complex and requires strategic planning, training and precise management.
The original standard of Internet Protocol, IPv4, was developed to handle addressing. Developed by the Defense Advanced Research Projects Agency (DARPA), IPv4 supports 32 bits of addressing for a total of 4,294,967,296 unique Internet addresses. At the time of IPv4’s development in 1970, the world population was less than 4 billion, and there were only a limited number of devices connecting to IP networks.
The DARPA government network was later managed by the National Science Foundation (NSF), and still later universities in the U.S. joined to create the Internet as we know it today. As technology evolved some years later, new services were born: Web browsers, e-mail, HyperText Markup Language (HTML) and e-commerce. The rapid explosion of Internet resources fueled the consumption of the IPv4 address space to a degree well beyond what the creators had envisioned.
To conserve IPv4 addresses, a variety of solutions have been developed. These include subnetting, variable length subnet masks (VLSM), the use of proxy servers and network address translation (NAT). As outlined below, these solutions shared a common objective of slowing consumption of IPv4 addresses but differed significantly in implementation.
• Subnetting divides a large network, such as a class A network (16 million host addresses), into multiple equal sub networks sharing a single IPv4 network.
• VLSM provides variable length masks that efficiently accommodate the need to create networks of various sizes. A point-to-point connection only needs two addresses, while a local area network (LAN) may need 300 addresses. VLSM resolved an inherent inefficiency in subnetting: Not all LANs are the same size, so the creation of equalsize networks underutilizes IPv4 addresses.
• Proxy servers allowed hosts to submit Web page requests to a “dual-homed” server – a server with two network cards – or a router. The proxy server would fetch the request for the client and then store the information for future requests. Only a single public address was needed for proxy server implementation.
• NAT was developed as a replacement for proxy servers as the use and size of LANs grew. Per Request for Comment 1918 (RFC 1918), a white paper developed by the Internet Engineering Task Force (IETF) in the 1980s, NAT is used to map private addresses to public addresses. Using NAT’s Port Address Translation (PAT) algorithm, multiple private addresses are able to share a single public address or are extended to share a range of public addresses.
POINT OF NO RETURN
While all of these solutions have bought time for the cable industry and others, we are at the point of no return. The Internet Assigned Numbers Authority (IANA), the entity responsible for global IP address designation, asserts that fewer than 2 percent of contiguous classful IPv4 addresses remain (/8s) and predicts IPv4 address exhaustion as early as February 2011. A Regional Internet Registry such as the American Registry for Internet Numbers (ARIN) is slightly more optimistic, predicting exhaustion of IPv4 addresses by the end of 2011. The dates listed for IANA and ARIN are variable since these dates are based on data over the past two years.
While those horizons are fast approaching, the problem of dwindling IPv4 addresses is even more imminent to the cable industry already. Let’s look at an example of an operator with 1 million basic video subscribers. Each set-top box uses an IP address for the control plane or to manage the STB using a protocol similar to IP’s Simple Network Management Protocol (SNMP). To support interactive services such as video-on-demand, real-time VoIP, instant messaging (IM) or Web browsing, an additional IP address may be required.
The support for services like the Multimedia over Coax Alliance (MoCA) protocol to share video content between STBs and customer premises equipment (CPE) in the home network requires additional addressing from the IP protocol.
Even without including current and future products like embedded multimedia terminal adapters (EMTAs), cable modems (CMs), embedded digital voice adapters (E-DVAs), e-Routers and wireless products, the consumption of IPv4 in this example is more than 2 million IP addresses. On average, cable operators are predicting consumption will exhaust the private 10 address space, creating a need for operators to draw on their public address space. Based on RFC 1918, cable operators could use all 273 private networks available; however, the network would be discontiguous.
IPv6 addressing provides the contiguous network solution operators need to scale the network for future services. In addition, IPv6 also will provide global routing scalability and rich new features for cable operators.
THE IPv6 SOLUTION
The Internet Engineering Task Force, an international standards organization that defines the way we use and manage the Internet, proposed the creation of IPv6 as early as 1994 to deal with the depletion of the IPv4 address space. In 1996, the IETF created 6bone to provide an IPv6 backbone for the testing of standards and implementations.
Because the United States has the majority of the capacity of the IPv4 address space (approximately 49.5 percent), countries such as Japan, China, the United Kingdom, Korea and Canada have been motivated to acquire additional addressing more quickly than has the U.S. The additional addressing has been needed to feed the rapid growth of countries like China. Because China has 338 million Internet users but only 8 percent of the IPv4 space, China led the world by creating the largest IPv6 network – the ChinaNet Next Carrying Network (CN2).
Currently there are 4 billion IPv4 addresses (2^32) with a world population of close to 6.9 billion. IPv6 is a 128-bit address system supporting 340 trillion trillion trillion (2^128), or 340 undecillion, addresses. That’s 49,534,488,380,470,-414,794,388,097,072 addresses per person on the planet today. IPv6 uses hexadecimal (hex), a 16-bit (2^4) numbering system: IPv6 addresses have 8 hextets, or sections, of 4 hex digits, and each hextet is made from 4 hex digits, equal to 16 bits per section. By comparison, the same address in IPv4 would need to be written in an unwieldy 128-digit format.
The address has a prefix, subnet portion and interface portion. The prefix defines the scope and/or type of address. For example, a 2000::/3 address represents a global unicast scope, a FE80::/10 represents a link local scope and an FC00::/7 is used for site traffic. The slash 3, 10 and 7 represent the number of bits in the prefix similar to Classless Inter-Domain Routing (CDIR) notation used to represent networks and prefixes in IPv4.
Cable system operators are assigned a 2000 with 30 to 48 bits in the prefix from ARIN. For example, a cable operator that is assigned a /32 from ARIN has 32 bits for creating sub networks and 64 bits to assign to customers. Because IPv6 is designed to support the Plug and Play (PnP) self configuration of IP-aware devices, 64 bits in the host are required to perform Stateless address auto-configuration (SLAAC) and Extended Unique Identifier 64 (EUI-64).
While the primary benefit of IPv6 is the ability to efficiently meet demand for an expanded market for Internet addresses, the flexibility of the standard provides additional benefits. These include:
• Added Layer 3 security using Internet Protocol Security (IPSec) – IPSec is integrated into the IPv6 communication model and into the header using the “next header” field, offering enhanced encryption and authentication mechanisms that were an add-on in the IPv4 environment. Moving away from ARPstyle broadcasts to a multicast discovery method using Secure NDP will provide another layer of security.
• Quality of service – Integration of “flow labels” for better traffic prioritization, assigning packets to a sequence for improved routing. The 8-bit field “traffic class” enables priority of an IPv6 datagram.
• More efficient routing and transport – By updating fragmentation and reassemble procedures, supporting a multi-level address hierarchy, IPv6 is able to reduce size of Internet routing tables and improve routing efficiency. The new, sleek header of 8 fields with a fixed 40 bytes provides a predictable packet. Moreover, the new header is less complex and more robust than the 14-field IPv4 header, removing IP header checksums. Header extensions were added that provide more efficient forwarding, fewer limits, future scalability and the simplified addition of new options to IPv6. In addition, IPv6 allows the use of larger “datagrams,” or network layer packets, of up to 4.2 GB.
• Anycast replaces broadcast – The address resolution protocol (ARP) was removed from IPv6. The routing protocol in IPv6 directs packets using the most straightforward path, regardless of the recipient’s location. This has the effect of expediting delivery and reducing unnecessary traffic on the network.
• Simplified provisioning and discovery – IPv6 features SLAAC, enabling IPv6 hosts to self-configure when connected to a routed IPv6 network, DHCPv6 stateful and DHCPv6 stateless for automatic address configuration, and DHCPv6 prefix delegation that configures the user’s router with the prefix to be used for each LAN. In addition, IPv6 includes new services such as neighbor discovery protocol (NDP) for discovering other computers or routers, a redesign of Internet Control Message Protocol (ICMP) for network communication, and enhancements to the routing protocols.
Special additions for mobility devices include roaming capabilities using a fixed IP address instead of the changing IPv4 address of the past. This will allow the mobile node to always be reachable when connected to a foreign link.
Finally, with IPv6 handling all routing functions, there will be no need for the historic role of network address translation (NAT). Instead of being needed to execute the conversion between private and public addresses, NAT will be required only as part of the migration process to IPv6.
While the features alone would make IPv6 attractive in any scenario, the dwindling number of IPv4 addresses lends increased urgency to the situation. It is critical that operators define their IPv6 migration strategy. North American cable operators take the following steps to be ahead of the migration curve:
• Move quickly to register an IPv6 prefix with ARIN.net. A domain name for IPv6 will be needed for name resolving.
• Use existing IPv4 address policies and RFC 5375 to guide IPv6 addressing for the hierarchical network (core, region, site, CMTSs and devices).
• Evaluate operations, network infrastructure and systems to determine the capabilities of IPv6 addressing. CableLabs’ DOCSIS 3.0, PacketCable and e-Router support the IPv6 stack.
• Conduct a feasibility study to objectively determine if systems perform adequately.
• Create a comprehensive IPv6 implementation approach, and a coordinated business plan for an evolutionary transition will reduce expenses.
• Determine where changes will occur in the network and assign priority to these findings.
• Collaborate with vendors to establish a pilot for testing IPv6 migration features such as tunneling, translation, pure v6, 6RD, DS-Lite and dual-stack. The pilot can provide insight to the operator on the relative merits of a core-to-edge IPv6 deployment versus an edge-to-core IPv6 deployment.
• Deploy IPv6 initially for management and operation of CPE devices in selected areas of the network. Test provisioning, DOCSIS, monitoring, back office and IPv6 features following the IPv4 operational model.
• Address other areas of concern, including: the development of accurate timelines, training plans, name resolution, CPE testing, enterprise network and how to maintain IPv4.
As I said, the transition to IPv6 is challenging, but the long-term benefits far outweigh the immediate difficulties. By following these steps, operators can simplify initial deployment of IPv6 in the network in a way that makes the process as painless as possible for themselves, their employees and, most importantly, their subscribers.