The Case for IPv6

The adoption of a technology or a protocol with such significant implications but to which it is not straightforward to tie immediate revenue is a complex process. This is particularly true early in the adoption process when the technology lacks a clear champion, be it service provider, enterprise, consumer, or even equipment vendor. Although technologists might be interested and even supportive, the marketing team, usually tasked with making the "case" for adoption and deployment, is faced with a "chicken and egg" situation. For this reason, it is important to make a clear distinction between "interest in IPv6" and "having a business case for IPv6."

Sprint showed technical interest in IPv6 early. It has been actively involved in the standardization and testing of IPv6 since 1997, both to gain experience with deploying and supporting IPv6 and to identify areas where the protocol still has problems that need to be solved. In support of 6bone, Sprint acquired 6bone IPv6 address space (3ffe:2900::/24) and overlaid an IPv6 test environment on top of its IPv4 SprintLink infrastructure with the help of generic routing encapsulation (GRE) tunnels. SprintLink IPv4 customers were offered the possibility to connect to the 6bone network at no additional cost.

Figure 5-1 depicts the topology of the IPv6 overlay deployed in the SprintLink network.

Figure 5-1 Sprint's TestbedNetwork

In 1998, Sprint's IPv6 evaluation network had 15 customers; in 1999 it had 40, and by the end of 2000 it had 110. Between 2000 and 2002 the environment had four more IPv6-capable points of presence (PoP) added: Brussels, Washington DC, San Jose, and New York City. During this period Sprint was turning up two to three customers per week. In 2000 and again in 2002 Sprint acquired global IPv6 address space (2001:440::/32) from ARIN. In May 2004, the number of IPv6 tunneled connections was 300 and the number of native connections was 2; however, the frequency of requests for IPv6 services slowed down considerably to roughly one per week. The customer base continued to expand to eventually reach 400 customers across the entire network shown in Figure 5-1. Sprint continues to offer IPv6 services over the original environment under the 2001:440::/32 prefix even after 6bone was retired on June 6, 2006 (http://www.6bone.net/). Sprint has seen interest in IPv6 service, especially native IPv6, increasing dramatically over

Figure 5-1 Sprint's TestbedNetwork the last 6 to 12 months, first driven by the OMB mandate, and then by ARIN's announcement on May 21, 2007.9

NOTE Sprint has been active in the IETF Next Generation Transition

(NGTrans) and IP Next Generation (IPng) working groups and co-authored RFC 2772, 6Bone Backbone Routing Guidelines.10 Sprint engaged in other IPv6 evaluation efforts outside of 6bone, as well. It provided engineering support to the Moonv6 project (http:// www.moonv6.org/).

The work done on deploying and operating this test environment provided Sprint with invaluable expertise that it could leverage in planning an IPv6 strategy and the IPv6-based service deployment. RFC 2772 did in fact highlight the multihoming challenges due to the constraints imposed on the IPv6 protocol by the address allocation policies. As detailed in Chapter 2, "IPv4 or IPv6—Myths and Realities," with the availability of provider independent address allocation, multihoming can be implemented in IPv6 the same way it is implemented in IPv4. However, IETF still has to find a more scalable solution for multihoming, regardless of the version of IP.

Sprint has actively participated in standards work and test environments in the context of the general principles of customer demand and broad commercial scalability required for the introduction of a new infrastructure protocol such as IPv6. It believes in customer preparation and scalability:

• Be prepared for customers: Sprint's customers will be deploying IPv6 for many different reasons. Sprint must complete its own deployment in order to effectively serve customer needs. The sooner the production deployment of IPv6 is in place, the better Sprint is prepared to assist customers with their deployments, both from a service availability

9. "ARIN Board Advises Internet Community on Migration to IPv6," http://www.arin.net/ announcements/20070521.html.

10. "SprintLink IPv6 Services; Overview," http://www.sprintv6.net/Sprintv6.html.

perspective and the consulting/managed services perspective. The same concepts apply to the internal planning of the IPv6 deployment. Sprint's IP core must be IPv6-enabled first in order to support IPv6 services enabled on Sprint's various infrastructures that offer specific services (mobile, VPN, and so on).

• Determine and improve behavior at scale: The sum total of IPv6 deployments today is nowhere near the scale of the global IPv4 networks, in either traffic level or number of networks/routes. Existent large-scale deployments of IPv6 are not open to the Internet. Until more large network providers begin ubiquitous deployments of IPv6 and open them to the global Internet, there will not be a wide enough environment to understand all aspects of IPv6's behavior at Internet scale, nor to wring out possible remaining issues that must be solved for the protocol to fully replace IPv4.

NOTE At the beginning of 2008, there were roughly 250,000 IPv4 Internet routes compared to around 1000 IPv6 routes. Comparison of Iv4 and IPv6 topology can be evaluated from the CAIDA website.11

• Prepare for innovation: IPv6 will provide new capabilities to networks and networked applications. Ultimately, these new capabilities will improve existing services while opening the door for new, innovative ones. While driven by other factors (customer demand, address exhaustion), early adoption provides Sprint with an environment in which it can pursue innovative ideas and enable customers to try out and improve them.

• Establish leadership: Early engagement in standardization and testing efforts helps identify Sprint as a leader in the industry.

• Guide vendors: Sprint started to work closely with vendors to ensure IPv6 readiness of products. It provided requirements for IPv6 features in products and performed trials of wireline and wireless equipment.

11. CAIDA, "Visualizing IPv4 Internet Topology at a Macroscopic Scale, Visualizing IPv6 AS-level Internet Topology 2008" http://www.caida.org/research/topology/as_core_network/ and http:// www.caida.org/research/topology/as_core_network/ipv6.xml.

In the end, however, the investigation of the protocol, the result of a relatively small investment, has to evolve to full commitment to the protocol adoption in order to have IPv6 deployed in production. This transition requires a business case. Because a service provider of Sprint's size covers several different markets, there might be multiple business cases for the adoption of IPv6, each specific to part of Sprint's business.

By mid-2007, two major drivers for IPv6 adoption in the U.S. service provider market emerged:

• Request for service from federal agencies: The OMB mandate requiring the infrastructures of civilian federal agencies to become IPv6 capable by June 2008 led to concrete demand for production-level IPv6 services. The requirements are coming not only from civilian federal agencies and through the Networx contract, but also from organizations working with or for the federal government. To maintain market share with these customers, service providers had to provide IPv6 services in time for the OMB-mandated deadline.

• IPv4 address shortage: In the case of wireless services, Sprint is seeing a change in the IP usage profile from short hold times of pooled addresses (casual data access) to longer hold times. These changes are driven by new applications requiring always-on connectivity and by new user habits. These changes dramatically decrease the ability to oversubscribe IP address resources for wireless devices. IPv6 represents the clean solution for addressing the growth in terms of number of subscribers, the demands of new applications, and the future Fixed Mobile Convergence (FMC).

These two drivers created two clear and distinct business cases for the adoption of IPv6 in two of Sprint's infrastructures: the wireline services infrastructure must be enhanced to support dedicated IPv6 Internet access and IPv6 VPNs, and the mobile services infrastructure must be upgraded to deliver information and services to mobile users over IPv6. The former business case had a clear timeline associated with its implementation, to become operational before June 2008 to help federal customers meet the OMB mandate requirements. In the absence of a killer IPv6-based application for the mobile services, the timeline for the latter business case is defined by Sprint.

The experience gained with the IPv6 evaluation network helps Sprint expand the business case for IPv6 in the context of the knowledge needs of federal agencies migrating to IPv6. Sprint can provide IPv6 planning, deployment, and operation consultancy services.

0 0

Post a comment