Overview Pv6 has become an nternet Request For Comment (RFC) in 1995 [1]. Pv6 was developed to solve a number of problems associated with Pv4. Solving address space limitation and lack of inherent security (authentication and integrity) of Pv4 were the primary motivators for Pv6. Mobile communications industry saw Pv6 as an essential feature. Pv6 was defined as the default P addressing scheme for Universal Mobile Telecommunications Service (UMTS) release-99. Adoption of Pv6 didn't follow the anticipated path. Among multiple reasons for this, probably the primary was industry's ability to overcome limited Pv4 address space with Network Address Translation (NAT). NAT has extended life-time of Pv4 substantially even though it has introduced a significant number of problems; well documented in [2]. As of October 2010, Pv6 adoption is dismal [3]. Only 2% of the top 1000 websites have WirelessE2E LLC October 2010 Page 1 oI 35 www.wirelesse2e.com whI Popr reachability over Pv6 [4]. On the other hand, common industry consensus shows that last Pv4 address will be allocated to an end user before the end of 2012 and nternet will officially run out of Pv4 addresses [5]. With this looming deadline, we anticipate Pv6 adoption will pick up steam. Unlike other nternet operators, Mobile Network Operators (MNOs) had to adopt with address crunch early on. Early standardization activity based on Pv6 in 1990s didn't lead to adoption. nstead MNOs had to rely on NAT. This long term experience with NAT will allow MNOs to adopt Pv6 much easier and faster than fixed network operators. n this paper, we will review the status of nternet addressing, adoption of Pv6, depletion of Pv4, multiple proposals for Pv6 transition. We will identify specific transition methods for mobile networks and compare their relative strengths. We will summarize our findings with a set of recommendations for MNOs on how they can embark upon eliminating this business risk. Internet Addressing Pv4 was published as an nternet Engineering Task Force (ETF) RFC (RFC-791) in 1981 [6]. t used a 32-bit addressing scheme, potentially allowing over 4 Billion addresses. Original design of Pv4 dating back to 1973-1974 assumed an 8-bit network and a 24-bit host address [7]. This allowed only 256 networks but in 1973 this number looked perfectly fine when there were a handful university and research facility sites connected. When the limitation of 256 was discovered to be too restrictive, class- WirelessE2E LLC October 2010 Page 2 oI 35 www.wirelesse2e.com Losr mojor nerwor| rronsirion noppeneJ on Jon. 1 sr 198S wnen Nerwor| Conrrol Prorocol (NCP) wos snur- Jown ro qive woy ro TCP/IP. whI Popr based addressing schema was adopted. This allowed class A addresses (8-bit network and 24-bit host address), class B addresses (16-bit network and 16-bit host address) and class C addresses (24-bit network and 8-bit host addresses). Based on this schema, the first half of Pv4 address space was identified as class A. Class B address space occupied of the available address space whereas class C addresses took 1/8. Class-based addressing schema led to a very sparse allocation. A number of companies, universities, research and military organizations jumped on the band-wagon and they obtained their class A address space (providing 16M host addresses). Among the owners of class A addresses, we can recognize some well known names of technology industry such as General Electric, BM, Xerox, HP, Apple, AT&T. Some other companies, not necessarily associated with networking technology such as Prudential Securities, Eli Lilly, Halliburton are also among a number of companies with their class A address spaces. By 1993, inefficiencies of class based address allocation scheme were obvious. Class B address space providing 65K hosts was too large for many enterprises whereas class C address space containing only 256 addresses was found to be too restrictive. n order to overcome these limitations Classless nter Domain Routing (CDR) was introduced [8]. CDR relied on Variable Length Subnet Masking (VLSM) and allowed subnets of varying sizes in terms of host addresses. Following figure from a dynamic webpage [5] for Pv4 address report maintained by Geoff Huston (Asia Pacific Network nformation Center (APNC)) summarizes the rate of Pv4 address allocation dating back to 1984. WirelessE2E LLC October 2010 Page 3 oI 35 www.wirelesse2e.com Sron[orJ Universiry rnor nos o closs-A ossiqnmenr (S6.0.0.0/8) rerurneJ ir boc| ro ARIN ro ossisr wirn IPv4 Jeplerion. Un[orrunorely Sron[orJ nos been on exceprion. whI Popr Between 1988 and 1994-1995, class A address allocations resulted in a significant expansion in allocated address space. With the introduction of CDR, this trend slowed down between 1995 and 2002-2003. Since then with the proliferation of nternet in every facet of daily life, expansion of allocated address space accelerated. As of October 2010, nternet Assigned Numbers Authority (ANA) has only 14 /8 address spaces left at its disposal. [5] predicts that by June 12 th 2011, ANA will allocate last of its available /8 addresses to a Regional nternet Registry (RR). Similarly, the same report predicts that February 2 nd 2012 will be the first date for an RR to run out of its available Pv4 address space. Once the Pv4 address space is depleted, a new commercial environment is expected to come to picture. Probably this would have positive benefits such as reducing the inefficiencies in address allocation within networks. On the flip side, it will also increase the cost of acquiring addresses and in turn increase the cost of using nternet. This last point WirelessE2E LLC October 2010 Page 4 oI 35 www.wirelesse2e.com Anorner 16 /8 oJJress spoces were iJenri[ieJ os closs L oJJresses mor|eJ [or [urure use. ILTF nos been resisronr ro re-ossiqn rnese oJJresses [or public or privore unicosr oJJressinq. Usinq rnem ro exponJ privore oJJressinq nos merirs. whI Popr and its unknown extent are the major motivations to migrate to Pv6. IP Addressing in MobiIe Networks Mobile communications industry identified the need for Pv6 as early as 1997-1998. Original Third Generation Partnership Program (3GPP) UMTS standards (release-99) were written with the assumption of Pv6 as the default packet data protocol and address type. Early enthusiasm of the mobile industry didn't translate into operational deployments. Certainly prime reason for this was the lack of Pv6 content. Furthermore, factors such as longer P packet headers, lack of suitable handset Operating System (OS), lack of Pv6 applications all contributed to mobile industry's abandonment of Pv6 and adoption of Pv4 with Network Address Port Translation (NAPT) 1 . MNOs' adoption of NAT along with Pv4 addressing for customers was mainly successful due to the following factors: 1. Data service adoption growth lagged the growth of mobile networks. Until General Packet Radio Service (GPRS) and One times Radio Transmission Technology (1xRTT) were introduced, data service was limited to modem emulation over Circuit Switched Data (CSD). Only after the introduction of 3G and more recently with smart phones such as iPhone, Android, Blackberry data service adoption accelerated. 2. nitially data services were primarily restricted to operator's walled garden. Wireless Application Protocol (WAP) and operator value- added services for content promoted this restriction. Handset to 1 We will reIer to NAPT as NAT in the rest oI the document unless we need to distinguish it Irom basic NAT. WirelessE2E LLC October 2010 Page 5 oI 35 www.wirelesse2e.com Oriqinolly UMTS releose 99 sronJorJs mor|eJ IPv4 os oprionol. Loc| o[ ovoiloble rerminol Jevices wirn IPv6 convinceJ SGPP ro cnonqe rnor by releose 4. whI Popr servers within walled-garden communication didn't need NAT. 3. Traffic volume for data usage was miniscule compared to traffic volumes operators started experiencing for the last 2-3 years since the introduction of smartphones. Low traffic volume lessened the impact on NAT devices. 4. nitial data devices were rarely always-on. That reduced the impact on NAT platforms in terms of the number of translations they needed to maintain. Few always-on devices like Blackberry used their dedicated private address scheme that allowed communication with cloud servers. Such communication didn't put any burden on carrier's NAT platforms. 5. Handset-to--handset communication applications were typically banned. nstead developers relied on application layer client- server model as opposed to peer-to-peer communication. 6. Mobile devices were designed and promoted to be clients as opposed to servers. For customers who need mobile servers, custom solutions such as public P addressing or tunneling based topologies were used. Even though deployment of NAT was a success, it introduced numerous compromises that resulted in various limitations in services provided to customers: 1. Standard applications like Hyper Text Transfer Protocol (HTTP), File Transfer Protocol (FTP), Simple Mail Transfer Protocol (SMTP) worked well behind NATs. On the other hand, more involved protocols like Real Time Streaming Protocol (RTSP), WirelessE2E LLC October 2010 Page 6 oI 35 www.wirelesse2e.com Unril recenrly mobile nerwor|s noJ very limireJ uplin| bonJwiJrn. Tnis reJuceJ rne incenrives ro builJ mobile servers or usinq mobile Jevices os conrenr Jonors [or peer-ro- peer opplicorions. whI Popr many Voice over P (VoP) flavors suffered even though NAT vendors worked hard to provide updates to their Application Level Gateway (ALG) functions. Many application developers migrated towards http transport. 2. Consumer use of mobile data networks is defined to be mobile client only model. Everything else was deemed to be vertical, needing different Access Point Names (APNs), devices, etc. 3. Operators didn't have the level of granularity to record usage at the network boundary and tie a particular session back to an individual customer. This impacted their ability to handle subpoena requests coming on behalf of other nternet providers looking for abusers among MNO customers. 4. NAT platform became a weak point in the network that was at risk to be attacked (denial-of-service) as well as impacting overall service in case of platform failures. 5. NAT complicated network design; requiring either co-location with gateway platforms or artificial routing methods in order to keep session consistency. 6. Even with the use of private P addressing, MNOs ran out of Pv4 (private) addresses. This required replicating the same RFC- 1918 address space numerous times or using BOGON-space. First option fragmented operator networks even more whereas the second option resulted in frequent re-numbering every time a previous BOGON was assigned. With the impending Pv4 exhaustion date, it is quite imperative for MNOs WirelessE2E LLC October 2010 Page 7 oI 35 www.wirelesse2e.com As on exomple Verizon re-uses rne enrire RFC-1918 oJJress spoce in 40 locorions [or NAT. whI Popr to start the Pv6 deployment effort. Considering the lack of Pv6 content and the established Pv4 legacy, Pv6 transition is a major effort that is expected to take a long period of time. This long transition is expected to require sizable capital and operational expenditure. Picking a wrong choice for transition may lead an MNO to burn unnecessary capital and lose valuable time. IPv6 Transition Techniques Pv6 transition has been in planning for the last 15 years. Throughout this period, numerous proposals have been developed to resolve various issues with transition. These proposals can be classified within three major classes: Dual-stack techniques are methods that rely on having both Pv4 and Pv6 capabilities in network nodes as well as hosts. This provides the most flexible transition strategy since hosts pick the right protocol stack based on the application requirements and network treats Pv4 and Pv6 equally. Tunneling techniques are methods that rely on encapsulating one type of network traffic over the other during the transition period. nitial motivation for tunneling techniques was to interconnect Pv6 islands through Pv4 networks. Translation techniques are methods that rely on converting protocol data units belonging to one protocol to another. Primarily they will be useful when transition needs to take place with the smallest impact to fewest number of nodes. WirelessE2E LLC October 2010 Page 8 oI 35 www.wirelesse2e.com whI Popr Before we look at the mobile network specific transition strategies, let us review the industry status, look at different classes of transition techniques and identify trends. DuaI Stack based Transition Methods Dual stack based transition is described as the preferred path per RFC- 4213 [9]. Based on this model, all network operators as well as end users that are currently running Pv4, must obtain an Pv6 prefix, enable Pv6 routing within their networks as well as interconnectivity with other networks and enable it on hosts. While Pv6 is being turned up, they should leave Pv4 stable until application quality and coverage parity is obtained between these two versions of P networking. When parity is achieved, economic justification for running Pv4 will disappear and Pv6 will become the sole networking protocol. Following figure shows the dual protocol stack for a host. Unfortunately the last 15 years have shown us that end users and network operators couldn't be motivated to invest in this dual stack strategy. Following figure from [5] shows the Autonomous Systems (AS) WirelessE2E LLC October 2010 Page 9 oI 35 www.wirelesse2e.com Lven rnouqn Juol- sroc| is JeemeJ ro be rne mosr [lexible ir su[[ers [rom per[ormonce concerns especiolly wnen IPv6 nerwor| moy nor be os relioble os IPv4 (simply Jue ro lesser operorionol orrenrion). whI Popr that are announcing their Pv6 prefixes on the nternet. t clearly identifies lingering pace of Pv6 adoption until 2009. Compared to AS that are advertising Pv6 prefixes, there are more than 35000 AS that are advertising Pv4 addresses as of October 2010. Certainly in this unbalanced situation of adoption, it is hard to follow through the original dual stack based transition plan since well before there is enough interesting Pv6 content, registries will run out of Pv4 addresses. This will require either Pv6 only hosts or dual stack hosts using private Pv4 addresses. A number of tunneling and translation based transition strategies were designed to handle the impact of private Pv4 addressing. We will review those in detail in the next two sections. TunneIing based Transition Methods During the last 15 years, many Pv6 transition methods using various forms of tunneling were proposed. Following is a short summary of some of these options: 4-in-6 was originally defined in RFC-2473 [10]. t assumes an WirelessE2E LLC October 2010 Page 10 oI 35 www.wirelesse2e.com Our o[ S500 or so AS, Fronce Telecom (FT) oJverrises rne snorresr IPv6 pre[ix (biqqesr oJJress spoce), o /19. Tnor equores ro 1/65000 o[ rne oJJress spoce currenrly ollocoreJ by IANA [or unicosr rro[[ic. whI Popr Pv6 transport network allowing Pv4 only hosts to communicate. As a transition method, it can be useful later stages of transition effort when majority of networks support Pv6 while there are Pv4 islands left. Since it is primarily based on static configuration of Pv6 tunnels and Pv6 only transport networks are not common, its adoption was very limited. Following figure shows the 4-in-6 topology. 6-in-4 was defined in RFC-4213 [9]. t assumes an Pv4 transport network allowing Pv6 only hosts to communicate. Tunnels are pre-configured and Pv6 traffic is encapsulated over Pv4 using a new protocol type (41). As a transition method, it is useful to interconnect Pv6 islands to each other over Pv4 nternet. t allows host-host, host-router and router-router configurations. Pre- configured nature of tunnels limits the applicability of this solution to router-router connections. Following figure shows the example topology for 6-in-4. WirelessE2E LLC October 2010 Page 11 oI 35 www.wirelesse2e.com Number o[ orqonizorions incluJinq Freener6, Hurricone Llecrric, SixXS operore public no-cosr 6-in- 4 runnel bro|ers ocross rne worlJ ro ollow experimenrorion. whI Popr 6-over-4 was defined in RFC-2529 [11]. t assumes an Pv4 transport network allowing Pv6 only hosts to communicate. Unlike 6-in-4, 6-over-4 tunnels can be dynamic. Tunnel end points are defined from Pv4 addresses by prepending a fixed pattern (fe80) to generate a link local Pv6 address. Similar to 6-in-4, it uses protocol type 41 for encapsulation. t is dependent on the use of Pv4 and Pv6 multicast implementation. Since inter-domain Pv4 multicast implementations are so rare, it is primarily applicable for allowing Pv6 hosts within a single domain. 6-to-4 was defined in RFC-3056 [12]. t assumes an Pv4 transport network allowing Pv6 islands to interconnect. Similar to 6-over-4, tunnels can be dynamic. Tunnel end points are defined from Pv4 addresses by prepending a fixed pattern (2002::/16) to generate a link local Pv6 address. Unlike 6-over-4, it doesn't need Pv4 multicast capabilities. Similar to 6-in-4 and 6-over-4, 6- to-4 is stymied with the lack of significant adoption of Pv6. 6-to-4 handles the shortage of Pv4 addresses well such that it needs only a single public Pv4 address per network to generate Pv6 addresses and it allows the interworking with NAT. Following diagram shows an example of 6-to-4 mechanism with relevant Pv6 subnet addressing based on the site Pv4 address. WirelessE2E LLC October 2010 Page 12 oI 35 www.wirelesse2e.com Currenrly rnere ore 40-50 AS onnouncinq 6ro4 pre[ix, proviJinq 6ro4 reloy service. (nrrp://www.qeripv6. in[o/inJex.pnp/Firsr _Sreps_[or_ISPs) A Since 6ro4 reloys nove ro occepr rro[[ic [rom ony nosr, incenrive ro moinroin quoliry service is low. Tnis leJ ro rne Jevelopmenr o[ 6rJ. whI Popr 6rd was defined in RFC-5969 [13]. t is an improvement of 6-to-4, where an nternet Service Provider (SP) specific prefix for the mapping between Pv4 and Pv6 local address can be used instead of the fixed prefix. This increases the incentive to deploy 6rd since the SP controls the particular border router for its customers. Furthermore, 6rd can be deployed when private Pv4 addresses are used within the SP domain. However, in this case NAT44 at the network border is still essential for Pv4 content. 6rd has received a lot of attention from SPs that are trying to jump- start Pv6 deployment. Free in France, Softbank in Japan, Comcast in the USA are among the pioneers of 6rd adopters. 6rd can be used as complementary to a dual stack host implementation. n this mode, it can be considered as an effective transition strategy in networks where Pv6 network node deployment is lagging. Following figure shows the potential deployment scenario for 6rd within a wired SP environment.
DS-Iite is currently being defined in ETF [14]. DS-lite is dependent on adoption of Pv6 as the networking technology WirelessE2E LLC October 2010 Page 13 oI 35 www.wirelesse2e.com Frencn ISP Free nos JeployeJ 6rJ ro proviJe IPv6 ro irs cusromers in 5 wee|s. Tnis exploins wny rJ sronJs [or "ropiJ Jeploymenr" os well os rne iniriols o[ irs creoror Remi Despres. whI Popr within an SP domain and allowing Pv4 traffic between dual stack hosts in customer premises and Carrier Grade NAT (CGN) devices at the SP boundary over Pv6 tunnels. t has a further refinement called gateway-initiated DS-lite [15] that allows various types of tunneling technologies, such as Generic Routing Encapsulation (GRE), Layer 2 Tunneling Protocol (L2TP), GPRS Tunneling Protocol (GTP) to map DS-lite tunnels. DS-lite along with a dual stack host implementation will reduce the need for double-NATting while permitting the native Pv6 traffic to reduce the overall load on the NAT44. n the original form of DS-lite, home gateway device implementing Basic Bridging BroadBand element (B4) function communicates with the NAT device implementing Address Family Transition Router element (AFTR) as shown in the following figure. DSMIPv6 was defined in RFC-5555 [16]. Dual Stack Mobile Pv6 allows a host (Mobile Node (MN)) to communicate with a Home Agent (HA) using MPv6 control protocol while allowing it to use both Pv4 and Pv6 Home Address (HoA) as well as Pv4 and Pv6 WirelessE2E LLC October 2010 Page 14 oI 35 www.wirelesse2e.com Anorner mojor bene[ir o[ DSMIPv6 is irs obiliry ro proviJe IP [low mobiliry berween WiFi onJ mocro wireless nerwor|s. whI Popr Care of Address (CoA). Furthermore, it allows NAT traversal capabilities in situations where MN is allocated a private Pv4 address. Primary penalty of DSMPv6 is the tunnel overhead for both Pv4 and Pv6 traffic. Following figure shows a MN can have either Pv4 or Pv6 tunnel back to its HA where the tunnel carries Pv4 or Pv6 traffic. There are a number of other methods that can be classified under tunneling such as Teredo [17], SATAP [18], Multi Protocol Label Switching (MPLS) Pv6 Provider Edge (6PE) [19]. Due to limited space of this document, these will not be further discussed in detail. Teredo and SATAP are methods that deal with Pv6 host-host (subnet-subnet) between Pv6 islands. Teredo specifically deals with NAT traversal for such tunnel traffic whereas SATAP handles dynamic tunnel management capability. Both of these mechanisms have been deployed in Microsoft OS but have seen limited adoption. MPLS 6PE solves the inter Pv6 island communication based upon advertising Pv6 routes learned from PE devices through MPLS. This reduces the impact on the WirelessE2E LLC October 2010 Page 15 oI 35 www.wirelesse2e.com Anorner MPLS mecnonism, 6VPL is Je[ineJ in RFC- 4659 j40]. Ir Jescribes usinq MPLS loyer S Virruol Privore Nerwor| (VPN) concepr [or roure Jisrriburion onJ ir ollows novinq virruol rourer insronces on PL rourers. Tnis woulJ ollow mulri- cusromer supporr over o corrier nerwor|. whI Popr rest of the core network elements. TransIation based Transition Methods Last group of transition methods are based upon the use of protocol translation between Pv4 and Pv6. A number of early protocol translation methods were developed within ETF around 1999-2000. Foundation for this work was the Stateless P/CMP Translation (ST) Algorithm defined in RFC-2765 [20]. ST was used as the basis for methods such as Bump n the Stack (BS) defined in RFC-2767 [21], Network Address Translation - Protocol Translation (NAT-PT) defined in RFC-2766 [22] and Bump n the AP (BA) defined in RFC-3338 [23]. None of these solutions have seen any significant deployment but NAT-PT was identified as a potential mechanism for Pv6 transition for 3GPP networks in RFC-4215 [24]. However, in 2007, ETF decided to deprecate NAT-PT due to numerous problems that have been documented in RFC-4966 [25]. However, this didn't eliminate the need for a transition solution that is using protocol translation. ETF formed a Working Group (WG), "behave, to identify a number of transition methods. Following are three of these that are refinements of the original NAT-PT method: NAT64/DNS64 is defined in ETF WG drafts [26],[27]. t is considered to be the stateful method of ETF protocol translation framework defined in [28]. NAT64 provides a mechanism for Pv6 only hosts to communicate with Pv4 only servers. Unlike the historical NAT-PT approach, it provides this by splitting NAT and DNS ALG function. Similar to dual stack options (using tunneling) like DS-lite, 6rd, it allows native Pv6 traffic from host to Pv6 server without any involvement of other network elements. Unlike WirelessE2E LLC October 2010 Page 16 oI 35 www.wirelesse2e.com ILTF nos olwoys been very resisronr ro usinq NAT. Mony RFCs were wrirren ro unJerline problems o[ NAT. One o[ rne mojor bene[irs promoreJ [or IPv6 wos rne obiliry ro eliminore NAT. Tnere[ore, ir is rorner inreresrinq ro see Inrerner Arcnirecrure BoorJ (IAB) ro semi- enJorse NAT [or IPv6 in RFC-5902 j41]. whI Popr DS-lite or 6rd, it doesn't need any access site intelligence, i.e., host and Pv6 access network and Pv4 server are completely unaffected by NAT64 functionality. Only devices that are needed are a CGN platform that provides NAT64 translation and a DNS64 platform mapping Pv4 addresses to NAT64 platforms. Because of this DNS flexibility, concerns with the historic NAT-PT solution such as single point of failure and scaling are not applicable for NAT64. Other major characteristic of NAT64 is it forces a clear break with Pv4 at the host since there is no Pv4 stack unlike the earlier dual stack solutions. This is an advantage to accelerate Pv6 deployment but as of 2010, there are still a number of applications that don't operate properly over Pv6. Some example applications are communications applications such as Skype, Gtalk, AM, multi-party games as well as websites that P address literals [29]. Same reference reports that for the top 1000 websites P address literal problem affects only 0.2% of the sites whereas for the top 10000 sites the failure rate goes up to 2%. Following diagram shows a mobile phone using an Pv6 address reaching to www.wirelesse2e.com that is currently hosted on an Pv4 server. Phone contacts the DNS64 for query. DNS64 reaches out to its authoritative DNS and then appends the Pv4 address with an Pv6 prefix associated with a particular NAT64 CGN device. NAT64 device strips the Pv6 prefix (in this example 2001:db8:8000::/96) and then uses the rest for Pv4 destination address. CGN's role on the Pv4 side is identical to a typical Port Address Translation (PAT) device. WirelessE2E LLC October 2010 Page 17 oI 35 www.wirelesse2e.com whI Popr
PNAT (Prefix NAT) s defined in ETF WG draft [30]. t was recently renamed as Bump n the Host (BH) method. As we discussed NAT64 cannot solve the connectivity requirements for Pv4 only applications since it doesn't have an Pv4 protocol stack. BH closes this gap by providing a translation from Pv4 to Pv6 within the host. BH provides DNS64 functionality of the NAT64 method by implementing an extension name resolver software module within the host. Similarly a NAT46 functional block translating Pv4 traffic to Pv6 within the host is used. BH suffers from the same problems NAT64 suffers when dealing with P address literals in protocol payload. IVI is an update of the ST RFC and defined in [31]. t is considered to be the stateless method of ETF protocol translation framework. Unlike the PAT method used in NAT64, V uses a one-to-one NAT translation. Practical limitation of this method is WirelessE2E LLC October 2010 Page 18 oI 35 www.wirelesse2e.com IVI (sronJinq [or IV: IPv4, VI: IPv6) wos implemenreJ in Cninese LJucorionol IP nerwor|. whI Popr the amount of public Pv4 addresses needed when the Pv4 addresses are near depletion. V is subject to same constraints NAT64 has when dealing with Pv6 only networks and hosts. Furthermore it suffers from the P literals similar to NAT64 and PNAT. A variant of V called double-V can be used to overcome the lack of Pv6 support for applications. n this mode, Pv4 Pv6 translation can be implemented in the host similar to what PNAT does. Proposed Methods for IPv6 Transition in MobiIe Networks 3GPP and ETF have been looking at Pv6 migration for a long time. As early as 2003, ETF has published informational RFC-3574 [32] analyzing the Pv6 transition scenarios. RFC-3574 covered GPRS transition as well as P Multimedia Subsystem (MS). t emphasized the fact that MS was designed as Pv6 only as of 3GPP release 5. RFC- 3574 was quite sparse in terms of the transition techniques that are applicable. nstead it enumerated all possible interworking needs and left transition techniques for ETF and 3GPP to identify. n 2005, ETF has developed informational RFC-4215 [24] mapping possible transition techniques to transition scenarios identified in RFC- 3574. RFC-4215 analyzed five transition scenarios for GPRS and two for MS. Let's look at the following two tables for GPRS and MS scenarios where we can review the relative applicability of the transition technique. WirelessE2E LLC October 2010 Page 19 oI 35 www.wirelesse2e.com whI Popr CPRS 1ransition scenario Proposed method Advantages Limitations Dual stack User Equipment (UE) connecting to IPv4 and IPv6 networks Deploy dual stack support in packet core. Use pre-conIigured tunneling when IPv6 Packet Data Protocol (PDP) context is not supported (e.g., roaming). It simpliIies network design. It doesn't impact IPv4 only applications. It doesn't solve NAT requirement (problems) Ior IPv4. It requires double packet core resources. It doesn't recommend a widely deployable solution Ior IPv6 over IPv4 tunneling. IPv6 UE connecting to IPv6 servers through an IPv4 network Use pre-conIigured tunneling between IPv6 islands over IPv4. It allows delaying upgrade oI operator backbone Ior IPv6 support. Numerous problems with managing tunnels and security issues associated with preconIigured tunnels. IPv4 UE connecting to IPv4 servers through an IPv6 network Use pre-conIigured tunneling between IPv4 islands over IPv6. This is not considered to be a likely scenario. IPv6 UE connecting to IPv4 servers Use NAT-PT in the operator network. Alternatively implement application level proxy devices. None. Better option is to stay with dual stack. NAT-PT has been deprecated. Application level proxies don't scale Ior multiple applications. IPv4 UE connecting to IPv6 servers Use application level proxy devices Ior limited number oI applications. This will happen when IPv6 adoption has taken place. Not an immediate problem. It allows mobile operator to keep IPv4 legacy without changes. Application level proxies can be built outside mobile operator network. Limited applicability. It doesn't solve the NAT requirement (problems) Ior IPv4. WirelessE2E LLC October 2010 Page 20 oI 35 www.wirelesse2e.com whI Popr IMS 1ransition scenario Proposed method Advantages Limitations UE connecting to an IPv4 node through IMS Deploy Back-to- Back User Agent (B2BUA) such as a Session Border Controller (SBC) between mobile IMS network and IPv4 network It allows interconnectivity to Internet based IMS hosts. Operator can extend IMS cloud to home, oIIice, etc. where IPv6 support doesn't exist. It breaks the end-to- end model. End-to- end security is broken. Two IPv6 IMS connected via an IPv4 network Use pre-conIigured tunneling between IPv6 islands over IPv4. It decouples transport network upgrades Irom IMS introduction. Numerous problems with managing tunnels and security issues associated with preconIigured tunnels.
More recently, 3GPP took up the task of identifying Pv6 transition scenarios. 3GPP TR 23.975 [33] describes a list of various scenarios considered for Pv6 transition. Similar to the churn within ETF with respect to Pv6 transition techniques, 3GPP seems to be struggling with having a large number of options and focusing on the most likely strategy to succeed. Considering many of the referenced ETF drafts have expired or merged with other proposals, this list needs to be updated once the official release 10 version of the TR becomes available. Most recent version (1.1.1) of TR 23.975 outlines the following options: A+P architecture is defined in an ETF draft [34]. Address+Port architecture is based on the fundamental recognition of depleted Pv4 address space and problems associated with port translation. A+P proposes to extend the Pv4 address space by overloading Transmission Control Protocol (TCP) or User Datagram Protocol (UDP) port numbers to identify end devices. t allows multiple WirelessE2E LLC October 2010 Page 21 oI 35 www.wirelesse2e.com AP orcnirecrure oJJresses on essenriol irem: li[e- spon onJ quoliry o[ NAT [or IPv4 neeJ ro be exrenJeJ ro compensore [or Jeloys in IPv6 rronsirion. whI Popr users to share the same Pv4 address similar to port-translation but without the need for a translation operation. Obviously this requires restricting the allowed ports per user to a subset such that the same P address can be multiplexed among multiple users. Mobile networks such as GPRS can be quite suitable such multiplexing since the logical tunnel end point device such as Gateway GPRS Support Node (GGSN) can be tasked to provide the Port Range Router (PRR) function. PRR provides routing traffic to and from hosts based on the A+P functionality. A+P architecture requires implementation in the UE and GGSN. t is orthogonal to the deployment of Pv6. n other words, it extends the lifespan of Pv4 for resolving NAT related problems and increasing the number of devices that can share a public Pv4 address. Therefore, it can be deployed in a network with Pv4 only PDP context. Following figure shows the use of A+P within a mobile network.
DS-Iite where Pv4 is tunneled over Pv6 to a CGN for NAT44 while Pv6 is carried natively. This transition model is applicable WirelessE2E LLC October 2010 Page 22 oI 35 www.wirelesse2e.com whI Popr for a UE with Pv6 PDP context only and both Pv4 and Pv6 protocol stacks. IVI where Pv4 traffic is translated to Pv6 in the UE and (stateless)-translated back to Pv4 at a CGN for Pv4 content while Pv6 is carried natively for UE. This transition model is applicable for a UE with Pv6 PDP context only and both Pv4 and Pv6 protocol stacks. DuaI-stack where Pv4 traffic is (port)-translated at a CGN for Pv4 content while Pv6 is carried natively for UE. This transition model is applicable for a UE with either two PDP contexts (one for Pv4 and another for Pv6) or an Pv4v6 PDP context and both Pv4 and Pv6 protocol stacks. PNAT/BIH where Pv4 traffic is translated to Pv6 in the UE and (port)-translated back to Pv4 at a CGN for Pv4 content while Pv6 is carried natively for UE. This transition model is applicable for a UE with Pv6 PDP context only and both Pv4 and Pv6 protocol stacks. NAT64/DNS64 where Pv6 traffic is (port)-translated to Pv4 at a CGN for Pv4 content while Pv6 is carried natively for UE. This transition model is applicable for a UE with Pv6 PDP context only and an Pv6 only protocol stack. Following table is a relative comparison of Pv6 transition strategies under consideration at 3GPP. WirelessE2E LLC October 2010 Page 23 oI 35 www.wirelesse2e.com whI Popr IPv 1ransition method Support for applications UE access UE capabilities Aetwork capabilities AP IPv4 applications need to accommodate limited number oI ports. Special handling Ior ICMP is needed. IPv4-only is adequate until IPv6 applications become essential. Implementing AP Iunctionality. AP PRR Iunction is needed in GGSN or P- GW. More public IPv4 addresses are needed compared to legacy port- translation. DS-lite IPv4 applications continue to be subject to NAT44 restrictions. No impact oI IP address literals. IPv6-only is adequate. Encapsulating IPv4 traIIic over IPv6. Implementing both IPv4 and IPv6 protocol stacks. NAT44 CGN and DS-lite termination capabilities. No need Ior private IPv4 address space. IVI IPv4 applications continue to be subject to NAT44 restrictions. Cannot handle IP address literals. IPv6-only is adequate. Implementing IVI Iunctionality. Implementing both IPv4 and IPv6 protocol stacks. NAT64 (stateless) CGN capability. No need Ior private IPv4 address space. Dual stack IPv4 applications continue to be subject to NAT44 restrictions. No impact oI IP address literals. Either IPv4v6 PDP or dual PDP (one Ior IPv4 and another Ior IPv6) Implementing both IPv4 and IPv6 protocol stacks. NAT44 CGN capability. No reduction in private IPv4 address space usage. PNAT/BIH IPv4 applications continue to be IPv6-only is adequate. Implementing PNAT/BIH Iunctionality. NAT64 (stateIul) capability. No WirelessE2E LLC October 2010 Page 24 oI 35 www.wirelesse2e.com whI Popr IPv 1ransition method Support for applications UE access UE capabilities Aetwork capabilities subject to NAT44 restrictions. Cannot handle IP address literals. Implementing both IPv4 and IPv6 protocol stacks. need Ior private IPv4 address space. NAT64/DNS 64 IPv4 applications using literals and some speciIic apps will Iail. IPv6-only is adequate. Implementing IPv6 protocol stack. NAT64 (stateIul) and DNS64 capabilities. No need Ior private IPv4 address space. Looking at the Pv6 transition methods, few observations can be made: Double-NAT is a more complex solution compared to single-NAT from end-to-end perspective. That rules out PNAT and (double) V methods. Since single V (NAT64 stateless) requires a public Pv4 address assignment for each mobile, it is not a realistic scenario. A+P reduces the incentive to migrate to Pv6 while requiring a significant capability upgrade in the UE. That substantial change as well as special treatment for CMP and fragment handling restrict the applicability of A+P. DS-lite provides a benefit of needing only an Pv6 PDP context as opposed to dual stack approach. Until release 8 upgrades take place relying on PDP type Pv4v6 is not possible. Other major benefit of DS-lite is the elimination for the private Pv4 address space. WirelessE2E LLC October 2010 Page 25 oI 35 www.wirelesse2e.com whI Popr For UE that doesn't have any Pv4 protocol stack, NAT64/DNS64 is the most viable option. t allows rapid transition to Pv6 while maintaining connectivity to Pv4 content servers via port- translation. How are Others Doing Transition? Mobile operators are not alone in being late for Pv6 transition. Across the entire communications industry, the level of readiness is quite poor. As we have demonstrated at the beginning of this paper, Pv6 adoption among content owners is seriously lacking. SPs that provide nternet service to large number of customers started sizable trials only in 2010. Similarly US government along with many other governments and organizations is quite late in completing meaningful Pv6 transition. Only recently, a new target of October 1 st 2012 was selected for Federal Agencies to have an Pv6 web presence. US government's efforts started much earlier. n February 2006, IPv6 Transition Guidance [35] was published by Federal CO Council Architecture and nfrastructure Committee. This document described the inter-agency efforts to accomplish the target of enabling network backbone Pv6 ready by June 2008. t provides a good blue-print for what needs to be included in a transition plan, objective and performance indicator setting as well as a governance structure to accomplish this transition. After achieving the June 2008 target for Pv6 backbone readiness demonstration, the Federal CO Council Architecture and nfrastructure Committee prepared and published another document titled "PIanning Guide/Roadmap Toward IPv6 Adoption within the US Government in May 2009 [36]. This document specified aggressive deadlines for Pv6 WirelessE2E LLC October 2010 Page 26 oI 35 www.wirelesse2e.com More criricol rorqer [or FeJerol Aqencies is Seprember S0, 2014 wnen rneir enrire inrernol in[rosrrucrure musr be upqroJeJ ro norive IPv6. whI Popr adoption and identified 2010 as the deadline for external facing servers to have Pv6 capability. The recent adoption of the new target date of October 2012 for the same milestone shows that US government is at least two years behind its original schedule. Another useful document on Pv6 transition planning is "A ProfiIe for IPv6 in the U.S. Government published by National nstitute of Standards and Technology (NST) in July 2008 [37]. t covers a set of general requirements that are applicable for all Pv6 nodes, as well as specific requirements for routers, network protection devices (firewalls, ntrusion Detection System (DS)/ntrusion Prevention System (PS)) and hosts. The documents specifies the test cases, and methodology. NST maintains an Pv6 validation program that uses accredited laboratories to test and certify equipment complying with the national Pv6 profile. Another major branch of US government taking a strong interest in Pv6 transition is the Department of Defense (DoD). DoD publishes "DoD IPv6 Standard ProfiIes for IPv6 CapabIe Products yearly. The most recent version, version 5.0 was published in July 2010 [38]. DoD Pv6 profile includes compliance requirements for all hosts, network appliances, servers, routers, switches and information assurance devices (firewalls, DS/PS, authentication servers, security gateways, Virtual Private Network (VPN) gateways and encryption devices). Profiles include mapping between various RFCs and listed equipment classes. Review of mentioned documents as well as comparable documentation from other bodies such as European Union, Australia, Canada show that many countries are behind schedule in their efforts but they also realize the need to accelerate them rapidly. Common industry target seems to WirelessE2E LLC October 2010 Page 27 oI 35 www.wirelesse2e.com As o[ Ocrober 2010, ICSA Lobs o[ Verizon, Universiry o[ New Hompsnire InrerOperobiliry Lob onJ Cnunq Hwo Telecom IPv6 Tesrinq Lob (Toiwon) ore rne only occreJireJ resr lobs [or equipmenr resrinq unJer rne NIST proqrom. whI Popr be 2012 for meaningful/sizable Pv6 adoption to occur. Preparing MobiIe Operator for IPv6 So far we have reviewed details of Pv6 transition for mobile networks from the perspective of methods and some of the technology changes in the UE and the network. Certainly migration towards Pv6 has a much bigger scope. n this section, we will highlight some of the other aspects of this transition effort and how mobile operators must prepare themselves. Transition planning documents prepared by organizations such US government Federal CO, NST, DoD must be used in parallel when mobile operators are going through their checklist for Pv6 transition planning. Mobile operator Pv6 transition plan must be developed through a series of steps that are relevant to operator's business priorities and objectives. Following can be considered as a generic set of steps to develop that transition plan: 1. Identifying business needs and objectives: For some operators providing Pv6 can be a strategic objective (early adaptor for differentiation, new machine-to-machine revenue opportunity, etc.). For others this could be an operational requirement (cost of NAT equipment, need for increases in service quality, etc.). 2. Identifying IPv6 enabIed services: Operators must decide what parts of their services will be Pv6 enabled. So far industry focus on enabling Pv6 has been UE and relevant customer facing applications. However, it might be possible to stretch this further WirelessE2E LLC October 2010 Page 28 oI 35 www.wirelesse2e.com whI Popr to include the entire mobile network including core, Radio Access Network (RAN), relevant transport nodes as well as enterprise network elements serving customer care, retail, finance, etc. applications for the operator. Obviously step 1 will determine the scope and extent of step 2. 3. Identifying technoIogy components: This step will be the foundational technology strategy development where particular transition technologies, types of equipment, hardware and software revisions, topology and configuration changes need to be decided. We will enumerate this step in further detail below. 4. DeveIoping transition strategy pIan: Federal CO Council document [36] provides an excellent blueprint for such a transition document. We encourage our readers to obtain it and analyze it in its entirety to develop their Pv6 transition plan. Borrowing from the referenced document, a mobile operator Pv6 transition plan must include priorities, activities, milestones, criteria for transition, dependencies, risks/mitigations, interoperability and security during the transition period, equipment Pv6 requirements, project governance, training and testing. We can add budgeting, procurement, field tests, friendly user trials to this list. 5. DeveIoping IPv6 market and enterprise introduction pIan: n addition to the technology transition plan, mobile operator must develop a customer (market) introduction and (if necessary) an enterprise introduction plan. Customer introduction plan must detail how Pv6 will be introduced to end users. Options include hiding the Pv6 change from the user (treating it as a technology- WirelessE2E LLC October 2010 Page 29 oI 35 www.wirelesse2e.com whI Popr driven event); promoting it for service differentiation; charging for a reduced service price for Pv6 for promotion. Technology components that need to be included in a mobile operator's Pv6 transition strategy must include the following items: Packet core changes: Based on the selected transition method, packet core network nodes must be upgraded. This includes release 8 support for GPRS nodes (Pv4v6 PDP context), any new APN settings, related DNS-G changes, validating roaming partner capabilities that support Pv6, any applicable GPRS Roaming eXchange (GRX) changes. f NAT needs to be integrated with GGSN or Long Term Evolution (LTE) P-GW due to the selected transition method, they have to be defined as part of the packet core changes. Core IP network changes: ndependent of the transition strategy, at least a portion of operator network must support Pv6. This requires building Pv6 islands, connecting them to each other via tunnels (such as MPLS 6PE or 6VPE), obtaining commercial Pv6 SP service for Pv6 nternet connectivity, (if necessary) establishing Border Gateway Protocol (BGP) peering with other Pv6 carriers, upgrading/replacing routers, switches, load balancers, upgrading NAT platforms for selected technology choice, upgrading information assurance devices (firewalls, DS/PS equipment, VPN), server platforms such as DNS infrastructure. Operator provided customer facing service changes: For operator's own customer facing services such as web portals, WirelessE2E LLC October 2010 Page 30 oI 35 www.wirelesse2e.com whI Popr content servers, self-help platforms, Multi-media Messaging System (MMS), etc. a parallel Pv6 strategy must be developed. Basic options include upgrading applications to a dual stack environment or relying on translation services from an intermediate platform such as load balancers. Strategy must also include how to include new services as opposed to legacy Pv4 based services. Upgrading other mobiIe network nodes: n order to support providing customers with Pv6 PDP context, other mobile network nodes such as Home Location Register (HLR), Policy Control elements may have to be upgraded or re-configured. Furthermore operator must decide what to do about MS, e.g., having a permanent Pv6 PDP context for MS, Pv6 connectivity for MS roaming, extending MS to nternet based clients using Pv4, etc. Upgrading business systems/processes: Existing provisioning, customer care, sales, and usage tracking systems may have dependencies on the P version. Telia-Sonera's Pv6 trial was reported to be blocked when problems with provisioning system was found. Such back-office systems may not have been designed parametrically and require upgrades beyond configuration changes. Integrating new IPv6 capabIe UE: Specification, validation, approval, deployment, support cycle for new user equipment must be executed numerous times. WirelessE2E LLC October 2010 Page 31 oI 35 www.wirelesse2e.com whI Popr ConcIusions Pv6 has been around for 15 years. However, its adoption in the nternet has been quite limited so far. Primary reason for the lagging adoption is the lack of economic incentive. Until now, end users or network operators haven't seen any reason to migrate to Pv6 since Pv4 addresses have been readily available. However, the situation is about to change. By 2012 the last Pv4 address will be assigned and possibly a new marketplace will start dictating the rules of scarcity. We believe this unpredictability will start driving Pv6 adoption quite rapidly in the next 2- 3 years. We are not as optimistic as American Registry for nternet Numbers (ARN) that was foreseeing to complete majority of migrations by 2012 [39]. Nevertheless, we expect sizable adoption (15-20% of global nternet traffic) to occur by 2012. This will create the impetus to drive even more migrations. Mobile networks is one of the primary reasons for transition towards Pv6. With over 4.6B customers, they will be by far the biggest constituent of Pv6. Even though 3GPP standardization for Pv6 has started long time ago, so far operational track record is non-existent. We anticipate this will change with the introduction of LTE when operators, especially those who are making a technology leap-frog, will grab the chance to make architectural upgrades such as Pv6 transition. Verizon's stance on making Pv6 mandatory LTE devices is a good example of this change. Mobile network operators must jump-start the Pv6 transition to be ready for introducing Pv6 UE by 2012 timeframe. This requires them to make a number of decisions rapidly. WirelessE2E LLC October 2010 Page 32 oI 35 www.wirelesse2e.com whI Popr We recommend the following for the Pv6 transition strategy: For service/UE introduction in 2011-2012 timeframe, rely on dual stack while testing and potentially deploying DS-lite solutions. This will avoid any impacts due to Pv4 only applications or P address literals in web content. For service/UE introduction beyond 2012, rely on single Pv6 stack while working with content and application providers to migrate them to Pv6. This will require deployment of NAT64/DNS64 solution for Pv4 content. Develop an Pv6 transition plan that would span 24-36 months. Start with small but quick wins such as building a test network, obtaining an Pv6 prefix, choosing and enabling a tunneling technology in the core P network, etc. in 4-6 months. Aim for a commercial UE with Pv6 capability in 24 months after project start. Limit network changes to essentials. As an example, instead of a dual-stack strategy for handset web portals (which are declining in importance for mobile operator), use a load balancer based translation method to reduce the amount of network and host changes. Upgrade packet core network to support release 8 as early as feasible. This will provide option for Pv4v6 PDP context. Evangelize change to your roaming partners, back-office, enterprise system owners so that they start taking action. WirelessE2E LLC October 2010 Page 33 oI 35 www.wirelesse2e.com whI Popr References |1| RFC-1883, 'Internet Protocol, version 6 (IPv6), December 1995. |2| RFC-2993, "Architectural Implications oI NAT", November 2000. |3| http://bgp.he.net/ipv6-progress-report.cgi |4| http://speedlab01.cmc.co.ndcwest.comcast.net:8088/monitor/ |5| IPv4 Address Report at http://www.potaroo.net/tools/ipv4/ |6| RFC-791, 'Internet Protocol, DARPA Internet Program Protocol SpeciIication, September 1981. |7| Vinton CerI, 'How the Internet Came to Be, as it appeared in "The Online User's Encyclopedia," by Bernard Aboba, Addison-Wesley, November 1993, ISBN 0-201-62214-9. |8| RFC-1518, 'An Architecture Ior IP Address Allocation with CIDR, September 1993. |9| RFC-4213, 'Basic Transition Mechanisms Ior IPv6 Hosts and Routers, October 2005. |10| RFC-2473, 'Generic Packet Tunneling in IPv6 SpeciIication, December 1998. |11| RFC-2529, 'Transmission oI IPv6 over IPv4 Domains without Explicit Tunnels, March 1999. |12| RFC-3056, 'Connection oI IPv6 Domains via IPv4 Clouds, February 2001. |13| RFC-5969, 'IPv6 Rapid Deployment on IPv4 InIrastructures (6rd)-Protocol SpeciIication, August 2010. |14| draIt-ietI-soItwire-dual-stack-lite-06, 'Dual-Stack Lite Broadband Deployments Following IPv4 Exhaustion, August 2010. |15| draIt-ietI-soItwire-gateway-init-ds-lite-01, 'Gateway Initiated Dual-Stack Lite Deployment, October 2010. |16| RFC-5555, 'Mobile IPv6 Support Ior Dual Stack Hosts and Routers, June 2009. |17| RFC-4380, 'Teredo: Tunneling IPv6 over UDP through Network Address Translations (NATs), February 2006. |18| RFC-5214, 'Intra-Site Automatic Tunnel Addressing Protocol (ISATAP), March 2008. |19| RFC-4798, 'Connecting IPv6 Islands over IPv4 MPLS Using IPv6 Provider Edge Routers (6PE), February 2007. |20| RFC-2765, 'Stateless IP/ICMP Translation Algorithm (SIIT), February 2000. |21| RFC-2767, 'Dual Stack Hosts using the "Bump-In-the-Stack" Technique (BIS), February 2000. |22| RFC-2766, 'Network Address Translation - Protocol Translation (NAT- PT), February 2000. WirelessE2E LLC October 2010 Page 34 oI 35 www.wirelesse2e.com whI Popr |23| RFC-3338, 'Dual Stack Hosts Using "Bump-in-the-API" (BIA), October 2002. |24| RFC-4215, 'Analysis on IPv6 Transition in Third Generation Partnership Project (3GPP) Networks, October 2005. |25| RFC-4966, 'Reasons to Move the Network Address Translator - Protocol Translator (NAT-PT) to Historic Status, July 2007. |26| draIt-ietI-behave-v6v4-xlate-stateIul-12, 'StateIul NAT64: Network Address and Protocol Translation Irom IPv6 Clients to IPv4 Servers, July 2010. |27| draIt-ietI-behave-dns64-11, 'DNS64: DNS extensions Ior Network Address Translation Irom IPv6 Clients to IPv4 Servers, October 2010. |28| draIt-ietI-behave-v6v4-Iramework-10, 'Framework Ior IPv4/IPv6 Translation, August 2010. |29| Jari Arkko, 'Experiences Irom an IPv6-Only World at Ericsson, Google IPv6 Implementors ConIerence, June 10-11, 2010. |30| draIt-ietI-behave-v4v6-bih-01, 'Dual Stack Hosts Using "Bump-in-the- Host" (BIH), October 2010. |31| draIt-ietI-behave-v6v4-xlate-23, 'IP/ICMP Translation Algorithm, September 2010. |32| RFC-3574, 'Transition Scenarios Ior 3GPP Networks, August 2003. |33| 3GPP TR 23.975: 'IPv6 Migration Guidelines, v 1.1.1, May 2010. |34| draIt-ymbk-aplusp-06, 'The AP Approach to the IPv4 Address Shortage, October 2010. |35| Federal CIO Council Architecture and InIrastructure Committee, 'IPv6 Transition Guidance, February 2006. |36| Federal CIO Council Architecture and InIrastructure Committee, 'Planning Guide/Roadmap Toward IPv6 Adoption within the US Government, May 2009. |37| National Institute oI Standards and Technology (NIST) special publication 500-267, 'A ProIile Ior IPv6 in the U.S. Government-Version 1.0, July 2008. |38| DoD InIormation Technology (IT) Standards Registry (DISR), 'DoD IPv6 Standard ProIiles Ior IPv6 Capable Products Version 5.0, July 2010. |39| RFC-5211, 'An Internet Transition Plan, July 2008. |40| RFC-4659, 'BGP-MPLS IP Virtual Private Network (VPN) Extension Ior IPv6 VPN, September 2006. |41| RFC-5902, 'IAB Thoughts on IPv6 Network Address Translation, July 2010. WirelessE2E LLC October 2010 Page 35 oI 35 www.wirelesse2e.com