idnits 2.17.1 draft-boucadair-behave-ipv6-portrange-03.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** The document seems to lack a License Notice according IETF Trust Provisions of 28 Dec 2009, Section 6.b.ii or Provisions of 12 Sep 2009 Section 6.b -- however, there's a paragraph with a matching beginning. Boilerplate error? (You're using the IETF Trust Provisions' Section 6.b License Notice from 12 Feb 2009 rather than one of the newer Notices. See https://trustee.ietf.org/license-info/.) Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- No issues found here. Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- ** There are 63 instances of too long lines in the document, the longest one being 10 characters in excess of 72. == There are 4 instances of lines with non-RFC6890-compliant IPv4 addresses in the document. If these are example addresses, they should be changed. == There are 8 instances of lines with non-RFC3849-compliant IPv6 addresses in the document. If these are example addresses, they should be changed. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust and authors Copyright Line does not match the current year == The document seems to contain a disclaimer for pre-RFC5378 work, but was first submitted on or after 10 November 2008. The disclaimer is usually necessary only for documents that revise or obsolete older RFCs, and that take significant amounts of text from those RFCs. If you can contact all authors of the source material and they are willing to grant the BCP78 rights to the IETF Trust, you can and should remove the disclaimer. Otherwise, the disclaimer is needed and you can ignore this comment. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (October 05, 2009) is 5289 days in the past. Is this intentional? Checking references for intended status: Informational ---------------------------------------------------------------------------- ** Obsolete normative reference: RFC 3315 (Obsoleted by RFC 8415) == Outdated reference: A later version (-04) exists of draft-bajko-pripaddrassign-01 == Outdated reference: A later version (-01) exists of draft-boucadair-dhcpv6-shared-address-option-00 == Outdated reference: A later version (-09) exists of draft-boucadair-pppext-portrange-option-01 == Outdated reference: A later version (-11) exists of draft-ietf-softwire-dual-stack-lite-01 Summary: 3 errors (**), 0 flaws (~~), 8 warnings (==), 1 comment (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 2 Network Working Group M. Boucadair, Ed. 3 Internet-Draft P. Levis 4 Intended status: Informational J-L. Grimault 5 Expires: April 8, 2010 A. Villefranque 6 M. Kassi-Lahlou 7 France Telecom 8 G. Bajko 9 Nokia 10 Y. Lee 11 Comcast 12 T. Melia 13 Alcatel-Lucent 14 October 05, 2009 16 Flexible IPv6 Migration Scenarios in the Context of IPv4 Address 17 Shortage 18 draft-boucadair-behave-ipv6-portrange-03 20 Status of this Memo 22 This Internet-Draft is submitted to IETF in full conformance with the 23 provisions of BCP 78 and BCP 79. This document may contain material 24 from IETF Documents or IETF Contributions published or made publicly 25 available before November 10, 2008. The person(s) controlling the 26 copyright in some of this material may not have granted the IETF 27 Trust the right to allow modifications of such material outside the 28 IETF Standards Process. Without obtaining an adequate license from 29 the person(s) controlling the copyright in such materials, this 30 document may not be modified outside the IETF Standards Process, and 31 derivative works of it may not be created outside the IETF Standards 32 Process, except to format it for publication as an RFC or to 33 translate it into languages other than English. 35 Internet-Drafts are working documents of the Internet Engineering 36 Task Force (IETF), its areas, and its working groups. Note that 37 other groups may also distribute working documents as Internet- 38 Drafts. 40 Internet-Drafts are draft documents valid for a maximum of six months 41 and may be updated, replaced, or obsoleted by other documents at any 42 time. It is inappropriate to use Internet-Drafts as reference 43 material or to cite them other than as "work in progress." 45 The list of current Internet-Drafts can be accessed at 46 http://www.ietf.org/ietf/1id-abstracts.txt. 48 The list of Internet-Draft Shadow Directories can be accessed at 49 http://www.ietf.org/shadow.html. 51 This Internet-Draft will expire on April 8, 2010. 53 Copyright Notice 55 Copyright (c) 2009 IETF Trust and the persons identified as the 56 document authors. All rights reserved. 58 This document is subject to BCP 78 and the IETF Trust's Legal 59 Provisions Relating to IETF Documents in effect on the date of 60 publication of this document (http://trustee.ietf.org/license-info). 61 Please review these documents carefully, as they describe your rights 62 and restrictions with respect to this document. 64 Abstract 66 This memo presents a solution to solve IPv4 address shortage and ease 67 IPv4-IPv6 interconnection. The document presents a set of 68 incremental steps for the deployment of IPv6 as a means to solve IPv4 69 address exhaustion. Stateless IPv4/IPv6 address mapping functions 70 are introduced and IPv4-IPv6 interconnection scenarios presented. 71 This memo advocates for a more proactive approach for the deployment 72 of IPv6 into operational networks. 74 This document provides the specification of the solution and 75 deployment scenarios together with IPv6 migrations paths. 77 Requirements Language 79 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 80 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 81 document are to be interpreted as described in RFC 2119 [RFC2119]. 83 Table of Contents 85 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 5 86 1.1. IPv4 Address Exhaustion . . . . . . . . . . . . . . . . . 5 87 1.2. To what extent does IPv6 solve the problem? . . . . . . . 5 88 1.3. Towards a proactive approach . . . . . . . . . . . . . . . 6 89 1.4. Contribution of this draft . . . . . . . . . . . . . . . . 7 90 1.5. Positioning this Draft . . . . . . . . . . . . . . . . . . 7 91 2. Terminology . . . . . . . . . . . . . . . . . . . . . . . . . 8 92 3. Reminder of the Port Range Solution . . . . . . . . . . . . . 9 93 4. IPv6-IPv4 Address Mapping Formalism . . . . . . . . . . . . . 10 94 4.1. IPv6 Prefix: Another IPv4-mapped Prefix Scheme . . . . . . 10 95 4.2. IPv4 to IPv6 Address Mapping Function . . . . . . . . . . 12 96 4.2.1. Overview . . . . . . . . . . . . . . . . . . . . . . . 13 97 4.2.2. Example . . . . . . . . . . . . . . . . . . . . . . . 13 98 4.3. IPv6 to IPv4 Address Mapping Function . . . . . . . . . . 13 99 4.3.1. Overview . . . . . . . . . . . . . . . . . . . . . . . 14 100 4.3.2. Example . . . . . . . . . . . . . . . . . . . . . . . 14 101 5. Stateless A+P Mapping Function . . . . . . . . . . . . . . . . 14 102 5.1. Stateless A+P Mapping gateway (SMAP) Function 103 description . . . . . . . . . . . . . . . . . . . . . . . 14 104 5.2. Implementation modes . . . . . . . . . . . . . . . . . . . 16 105 5.2.1. SMAP to route incoming traffic destined to a 106 shared IPv4 address . . . . . . . . . . . . . . . . . 16 107 5.2.2. No IPv4 capabilities are used anymore between two 108 SMAP-enabled nodes . . . . . . . . . . . . . . . . . . 17 109 5.3. Deployment Scenarios . . . . . . . . . . . . . . . . . . . 17 110 5.4. SMAP and PRR . . . . . . . . . . . . . . . . . . . . . . . 19 111 6. IPv6 Migration Scenarios . . . . . . . . . . . . . . . . . . . 19 112 6.1. Overview . . . . . . . . . . . . . . . . . . . . . . . . . 19 113 6.2. IPv6 Prefixes and Addresses . . . . . . . . . . . . . . . 20 114 6.3. On Stateless and Binding Table Modes . . . . . . . . . . . 21 115 6.3.1. Stateless Mode . . . . . . . . . . . . . . . . . . . . 21 116 6.3.2. Binding Table Mode . . . . . . . . . . . . . . . . . . 22 117 6.4. Step_0: IPv6 at Access Network . . . . . . . . . . . . . . 22 118 6.4.1. Context . . . . . . . . . . . . . . . . . . . . . . . 23 119 6.4.2. Overall Procedure . . . . . . . . . . . . . . . . . . 23 120 6.4.3. Focus on Communication Establishment . . . . . . . . . 26 121 6.4.4. Typical Flow Example . . . . . . . . . . . . . . . . . 27 122 6.5. Step_1: IPv6 a Means to Transfer Incoming IPv4 Packets . . 28 123 6.5.1. Context . . . . . . . . . . . . . . . . . . . . . . . 28 124 6.5.2. Overall Procedure . . . . . . . . . . . . . . . . . . 28 125 6.6. Step_2: Only IPv6 Is Used For Both Incoming and 126 Outgoing Traffic . . . . . . . . . . . . . . . . . . . . . 31 127 6.6.1. Context . . . . . . . . . . . . . . . . . . . . . . . 31 128 6.6.2. The IPv6 Encapsulation-Based Mode . . . . . . . . . . 32 129 6.6.3. The IPv6 Translation-Based Mode . . . . . . . . . . . 37 130 6.7. Analysis and IPv6 Migration Scenarios . . . . . . . . . . 42 132 7. IANA Considerations . . . . . . . . . . . . . . . . . . . . . 45 133 8. Security Considerations . . . . . . . . . . . . . . . . . . . 45 134 9. Acknowledgements . . . . . . . . . . . . . . . . . . . . . . . 45 135 10. References . . . . . . . . . . . . . . . . . . . . . . . . . . 46 136 10.1. Normative References . . . . . . . . . . . . . . . . . . . 46 137 10.2. Informative References . . . . . . . . . . . . . . . . . . 46 138 Authors' Addresses . . . . . . . . . . . . . . . . . . . . . . . . 47 140 1. Introduction 142 1.1. IPv4 Address Exhaustion 144 It is commonly agreed by the Internet community that the exhaustion 145 of public IPv4 addresses is an ineluctable fact. Regular alarms and 146 reports have been emitted by the IETF particularly by the reports 147 presented within the GROW working group (Global Routing Operations 148 Working Group) meetings. 150 G. Huston introduced an extrapolation model to forecast the 151 exhaustion date of IPv4 addresses managed by IANA. This effort 152 indicates that if the current tendency of consumption continues at 153 the same pace, IPv4 addresses exhaustion of IANA's pool would occur 154 in 2011, while RIRs'pool would be exhausted in mid-2012. The state 155 of the current consumption of public IPv4 addresses is daily updated 156 and is available at this URL: 157 http://www.potaroo.net/tools/ipv4/index.html. 159 1.2. To what extent does IPv6 solve the problem? 161 In this context, the community was mobilized in the past to adopt a 162 promising solution (in particular with the definition of IPv6). IPv6 163 has been introduced for several years as the next version of the IP 164 protocol. This new version offers an abundance of IP addresses as 165 well as several enhancements compared to IPv4. IPv6 specifications 166 are mature and current work within the IETF is related to operational 167 aspects. Despite this effort, IPv6 is not globally activated by 168 service providers for both financial and strategic reasons. 170 However, even if a service provider activates IPv6, it will be 171 confronted with the problem to ensure a global connectivity towards 172 nowadays Internet v4. Mechanisms such as NAT-PT (Network Address 173 Translation Protocol Translation) were introduced to ensure the 174 interconnection between two heterogeneous realms (i.e. IPv4/IPv6) 175 and to ensure a continuity of IP communications (i.e. End-to-end). 176 These solutions are stateful and are not suitable to interconnect an 177 IPv6 domain with a dominant Internet which is IPv4-only. Further 178 work should be undertaken with IETF to elaborate lightweight and 179 hopefully stateless solution to ease IPv4-IPv6 interconnection. 180 Moreover, Service providers should adopt clear strategies so as to 181 ease the adoption of IPv6 and to decrease the complexity related to 182 IPv4-IPv6 interconnection which is one of the critical issues to be 183 taken into account when designing service platforms. 185 Last, it is worth to mention that migrating to IPv6 is a service 186 provider issue and not the one of its customers. The ultimate 187 requirement of the customers (mainly residential and mass market) is 188 to benefit from a global IP connectivity. How this connectivity is 189 engineered and put into effect is of the business of the IP 190 connectivity service providers. Of course, some corporate customers 191 would specify the nature of their IP connectivity and reduce the 192 interconnection engineering complexity of their interconnection nodes 193 with the domain of their IP connectivity service provider(s). From 194 this standpoint, service providers should be more proactive in order 195 to avoid a crisis scenario where no IP addresses are available to be 196 assigned to their customers. 198 1.3. Towards a proactive approach 200 The introduction of IPv6 into public networks becomes a reality. 201 Several Internet providers have enabled IPv6 in their routers and 202 launched therefore their IPv6 migration operations. The portion of 203 the IPv6-enabled routers differs between SPs. The current trade is 204 that operators offer dual connectivity to their customers, i.e. IPv4 205 and IPv6 access. IPv4 connectivity usage should be gradually 206 decreased in favour of IPv6 one. This convergence phase towards a 207 pure IPv6 connectivity will take several years depending on the 208 policies adopted by service providers. For operators that adopt an 209 aggressive position with regards to the activation of IPv6, this 210 transition phase could be small compared to passive operators. 211 Nevertheless the overall Internet IPv6 coverage will be long. This 212 is due mainly to the significant number of involved actors to be 213 convinced for a full migration towards IPv6, the significant number 214 of existing ASes (more than 30000), etc. Moreover, customers do not 215 have any reason to modify their local architecture (e.g., a given 216 organisation does not have any motivations to migrate its FTP or HTTP 217 servers towards IPv6). Operators must expect a long work of 218 accompaniment for the migration towards IPv6. The final migration 219 towards IPv6 would take several years (at least 10 years). 221 This migration to IPv6 should be incremental and not implemented in 222 one shot. For these reasons, service providers should elaborate 223 migration scenarios so as to achieve a transparent migration. This 224 transparency is required because end-users should not be aware on the 225 underlying technology used to deliver their subscribed services and 226 the complexity related to service engineering should be hidden to 227 end-users. Furthermore, service providers should use means to 228 prioritise IPv6 traffic and the invocation of IPv6 transfer 229 capabilities without relying on end-users behaviour. IPv6 transfer 230 capabilities should be exploited and not considered as dormant ones. 231 If no proactive means/procedures are adopted, the ratio of IPv6 232 traffic will depend on the behaviour of end-users and also on 233 available IPv6 services. 235 Furthermore, in the perspective of IPv6 migration, the maintenance 236 and the operating of dual connectivity infrastructure would therefore 237 be required for a long period. This option is not to be encouraged 238 within service providers since it does not optimise both OPEX/CAPEX. 239 Both technical skills should be maintained within each individual 240 organisation. As an alternative, this document proposes a proactive 241 and incremental deployment approach which consists at: 243 - Activation of IPv6 and port range IPv4 solution at the same time. 244 Port-restricted devices are provisioned with an IPv6 prefix, a shared 245 IPv4 public address and a port range. 247 - Activation of stateless functions and use of IPv6 to carry IPv4 248 traffic from/to port-restricted devices. 250 - Migration to IPv6-only core network. 252 - Maintain stateless IPv4-IPv6 interconnection functions at 253 interconnection segment to not alter intra-domain paths. 255 1.4. Contribution of this draft 257 This memo defines several solutions to solve the IPv4 address 258 shortage problem and to migrate to IPv6 without requiring stateful 259 nodes. It proposes also several migration paths. This target IPv6 260 deployment is a long term objective and can be reached incrementally 261 through one or several intermediate steps. These intermediate steps 262 perimeters differ from a service provider to another one depending on 263 the service opportunities targeted by enabling IPv6-related 264 capabilities and also on the level of risks on the already running 265 (IPv4-based) services. Risks on existing services should be 266 assessed. Careful or aggressive position may be adopted by each 267 service provider. Service providers are free to deploy the step/the 268 migration path suitable to their context and objectives, etc. This 269 document only sketches scenarios and interconnection configurations. 270 Voluntarily, no frozen architecture is described. Several options 271 are supported. 273 This document provides: 275 o solution specification 277 o and deployment scenarios together with migrations paths. 279 1.5. Positioning this Draft 281 This draft proposes a solution to solve both IPv4 address exhaustion 282 and IPv4-IPv6 interconnection. Unlike 283 [I-D.ietf-softwire-dual-stack-lite], no additional NAT is required to 284 be deployed at the service provider's network. Both encapsulation 285 and translation modes are presented in this memo. 287 A set of issues related to IPv4 Internet access in the context of 288 public IPv4 address exhaustion are identified and described in 289 [I-D.levis-behave-ipv4-shortage-framework]. 291 2. Terminology 293 Within the context of this draft, the following terminology is used: 295 o Access segment: This segment encloses both IP access and backhaul 296 network. Within this document, this access segment encompass an 297 access PoP. 299 o Core network: Denotes a set of IP networking capabilities and 300 resources which are between the interconnection and the access 301 segments. 303 o Interconnection segment: Includes all nodes and resources which 304 are deployed at the border of a given AS (Autonomous System) a la 305 BGP (Border Gateway Protocol). 307 o PRR: Stands for Port Range Router. This function is responsible 308 to handle a port-based routing. This function may retrieve the 309 port value and use it to determine which routing action is to be 310 executed or use it together with the destination IP address to 311 build an IPv6 address. 313 o Interconnection PRR (i-PRR): A PRR which is deployed at 314 interconnection segment. 316 o Access PRR (a-PRR): A PRR which is deployed at access segment. 318 o SMAP: Stands for Stateless A+P Mapping. This function is 319 responsible to encapsulate (Resp. decapsulate), in a stateless 320 scheme, IPv4 packets in (Resp. from) IPv6 ones. A SMAP function 321 may be hosted in a PRR, end-user device, etc. 323 o Port-restricted device (PRD): A device which is able to constrain 324 its source port number to be within a given Port Range. A port 325 restricted device may be of several types such as: 327 * CPE (Customer Premise Equipment)/HGW (Home Gateway) 329 * PDA (Personal Digital Assistant) 330 * Mobile terminal 332 3. Reminder of the Port Range Solution 334 This section is a reminder of the solution presented in 335 [I-D.boucadair-port-range]. 337 The main principle of the solution is to assign the same IP address 338 (called Primary IP Address) to several end-users' devices and to 339 constraint the source port numbers to be used by each device. In 340 addition to the assigned IP address to access IP connectivity 341 services, an additional parameter, called Port Range, is also 342 assigned to the customer's device. For outbound communications, a 343 given port restricted device proceeds to its classical operations 344 except the constraint to control the source port number assignment so 345 as to be within the Port Range. The traffic is then routed without 346 any modification inside the Service Provider's domain and delivered 347 to its final destination. For inbound communications, the traffic is 348 trapped by a dedicated function called: Port Range Router (PRR). 349 This function may be embedded in current routers or hosted by new 350 nodes to be integrated in the IP infrastructure of these service 351 providers. Appropriate routing tuning policies are enforced so as to 352 drive the inbound traffic to cross a PRR. Each PRR correlate the 353 Primary IP Address and information about the allowed port values with 354 a specific identifier called: routing identifier (e.g., Private IPv4 355 address, IPv6 address, point-to-point link identifier, etc). This 356 routing identifier is used to route the packets to the suitable 357 device among all those having the same IP address. 359 Port-restricted devices are provisioned with port range to be used, 360 especially the Port Mask to be applied when selecting a port value as 361 a source port. A Port Mask defines a set of ports that all have in 362 common a subset of pre-positioned bits. This set of ports is also 363 called Port Range. Two port numbers are said to belong to the same 364 Port Range if and only if, they have the same Port Mask. A Port Mask 365 is composed of a Port Range Value and a Port Range Mask. 367 o The Port Range Value indicates the value of the significant bits 368 of the Port Mask. The Port Range Value is coded as follows: 370 * The significant bits may take a value of 0 or 1. 372 * All the other bits (non significant ones) are set to 0. 374 o The Port Range Mask indicates, by the bit(s) set to 1, the 375 position of the significant bits of the Port Range Value. 377 0 1 378 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 379 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 380 |1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0| Port Range Mask 381 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 382 | | | 383 | | | (3 significant bits) 384 v v v 385 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 386 |0 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0| Port Range Value 387 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 389 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 390 |0 0 1 x x x x x x x x x x x x x| Usable ports (x may take a value of 0 or 1). 391 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ 393 Figure 1: Example of Port Range Mask and Port Range Value 395 An example of port range is provided in Figure 1. Ports belonging to 396 this port range must have the 1st bit (Resp. the 2nd and 3rd), from 397 the left, set to 0 (Resp. 0 and 1). The Port Mask is represented as: 398 001xxxxxxxxxxxxx. 400 4. IPv6-IPv4 Address Mapping Formalism 402 This section discusses issues related to the building of an IPv6 403 prefix or IPv6 address using IPv4-related information: 405 4.1. IPv6 Prefix: Another IPv4-mapped Prefix Scheme 407 [I-D.boucadair-port-range] proposes to assign the same public IPv4 408 address together with a port range to several devices. In order to 409 discriminate those devices, an additional identifier called, routing 410 identifier, is required. This identifier may be a secondary IPv4 411 address, PPP session identifier, etc. This document assumes that 412 this identifier is an IPv6 address. This prefix is built using IPv4- 413 related information as illustrated in Figure 2. 415 +------------------------+----------+---------+ 416 | Pref6 | @IPv4 | PRM | 417 +------------------------+----------+---------+ 418 Max. 419 <-----------n bits------> < 32 bits> <16 bits > 420 <-----------------64 bits max.----------------> 422 Figure 2: IPv6 prefix enclosing an IPv4 address and a port range 424 1. The length of this prefix is recommended to be less than 64 bits; 426 2. Pref6: is a sub-prefix belonging to the service provider or well- 427 known prefix allocated by IANA for this service. The length of 428 this field is variable (may be different from a service provider 429 to another if not allocated by IANA). The Pref6 prefix used to 430 build an IPv4-mapped IPv6 prefix may not be the same as the one 431 used to execute the IPv4-to-IPv6 mapping function introduced in 432 Section 4.2. 434 3. @IPv4 field encloses the shared IPv4 address. The length of this 435 field is 32 bits; 437 4. PRM field includes the value of the significant bits of the Port 438 Range. The maximum length of this field is 16 bits. But, in 439 deployment scenarios this field may be 3, 4 or 5 bits. If n bits 440 are used to build the PRM, the same IPv4 address may be shared 441 between 2^n port-restricted devices. 443 For illustration purposes two examples are provided below. 445 Let suppose that a service provider dedicates the 2a01:c0a8::/29 to 446 build an IPv4-inferred IPv6 prefix. In this example, we suppose that 447 8 port restricted devices share the same public address 448 193.51.145.206 owing to a port range mask with three significant bits 449 (i.e. the three first bits are used to build the port mask. The 450 remaining 13 bits may take a 0 or 1 value ), yielding 8192 (2^16/2^3) 451 possible ports per each port-restricted device. The corresponding 452 IPv6Pref prefixes for these 8 port-restricted devices are thus the 453 following ones: 455 - 1st port-restricted device (Port Mask: 000xxxxxxxxxxxxx): 457 IPv6Pref = 2a01:c0a 1 11000001001100111001000111001110 000 :: 458 --------193.51.145.206---------- PRM 460 IPv6Pref = 2a01:c0aE:099C:8E70::/64 462 - 2nd port-restricted device (Port Mask: 001xxxxxxxxxxxxx): 464 IPv6Pref = 2a01:c0a 1 11000001001100111001000111001110 001 :: 465 --------193.51.145.206---------- PRM 467 IPv6Pref = 2a01:c0aE:099C:8E71::/64 469 - ... 471 - 8th port-restricted device (Port Mask: 111xxxxxxxxxxxxx): 473 IPv6Pref = 2a01:c0a 1 11000001001100111001000111001110 111 :: 474 --------193.51.145.206---------- PRM 476 IPv6Pref = 2a01:c0aE:099C:8E77::/64 478 In this second example, let suppose that the service provider 479 dedicates the 2a01:c::/20 prefix to build an IPv4-mapped IPv6 prefix. 480 If we consider that 193.51.145.206 address is shared between 16 (2^4, 481 4 bits are used as the significant bits of the port range) port- 482 restricted devices. The 16 port-restricted devices sharing that 483 address have the following IPv6Pref prefixes (only the first prefix 484 is represented below): 486 - 1st port-restricted device (Port Mask: 0000xxxxxxxxxxxx): 488 IPv6Pref = 2a01:c 11000001001100111001000111001110 0000 :: 489 --------193.51.145.206---------- PRM- 491 IPv6Pref = 2a01:cc13:391c:e0::/56 493 - ... 495 4.2. IPv4 to IPv6 Address Mapping Function 496 4.2.1. Overview 498 Within this memo, IPv4-to-IPv6 address mapping function denotes a 499 function which uses IPv4-related information, as conveyed in a 500 received IPv4 packet, to generate IPv6 one. This function generates 501 an IPv6 address which builds as illustrated in Figure 3. 503 o Pref6 is configured by the service provider. 505 o Then, the next 32 bits are set to the value of the destination 506 IPv4 address; 508 o The next 16 bits are set to the value of the destination port 509 number; 511 o The remaining bits are then set to zeros. 513 +----------------------------------------------------------------------...+ 514 | Pref6 |Dest. IPv4 |Dest. | 0:0:0:0 | 515 | |address |port | | 516 +----------------------------------------------------------------------...+ 518 Figure 3: Pref6A Address Format 520 4.2.2. Example 522 Let suppose that a given device is provided with the Pref6 prefix 523 equal to 2a01:c0a8::/29. Then the corresponding IPv6 address, using 524 the IPv4-to-IPv6 address mapping function, to the IPv4 address equal 525 to 193.51.145.206 and the port number equal to 19039 526 (0100101001011111) is the following: 528 IPv6PrefA=2a01:c0a 1 11000001001100111001000111001110 0100101001011111 :: 529 --------193.51.145.206--------- -----port------- 531 IPv6PrefA=2a01:c0aE:099C:8E72:52F8::/128 533 This IPv6 address falls in the IPv6 prefix of the second port- 534 restricted device (port range 001) as listed in the previous section. 536 4.3. IPv6 to IPv4 Address Mapping Function 537 4.3.1. Overview 539 Unlike the previous function, the IPv6-to-IPv4 address mapping 540 functions generates an IPv4 address together with a port number from 541 the header and the transport part of a received IPv6 packet as 542 follows: 544 o The destination IPv4 address corresponds to the 32 bits which 545 follow a per-configured Provider prefix; 547 o Destination port number is equal to the one of the received IPv6 548 packet. 550 4.3.2. Example 552 Let suppose that the Pref6 prefix equal to 2a01:c0a8::/29 is used. 553 Then the corresponding IPv4 address, resulting from the IPv6-to-IPv4 554 address mapping function applied to the address 2a01:c0aE:099C:8E71: 555 A5F8::/128 is 193.51.145.206 since: 557 2a01:c0aE:099C:8E71:A5F8 = 559 2a01:c0a 1 11000001001100111001000111001110 0100101001011111 :: 560 --------193.51.145.206--------- -----port------- 562 5. Stateless A+P Mapping Function 564 5.1. Stateless A+P Mapping gateway (SMAP) Function description 566 Stateless A+P Mapping gateway (SMAP) consists in two basic functions 567 as described in Figure 4. 569 1. SMAP encapsulates an IPv4 packet, destined to a shared IPv4 570 address, in IPv6 one. The IPv6 source address is constructed 571 (see Section 4.2) from the IPv4 source address and port number 572 plus the IPv6 prefix which has been provisioned to the node 573 performing the SMAP function. The destination IPv6 address is 574 constructed using the shared IPv4 destination address and port 575 number plus the IPv6 prefix which has been provisioned to the 576 SMAP function and which is dedicated to IPv4 destination 577 addresses. 579 2. SMAP extracts IPv4 incoming packets from IPv6 incoming ones which 580 have IPv6 source addresses belonging to the prefix of the node 581 performing the SMAP function. Extracted IPv4 packets are then 582 forwarded to the point identified by the IPv4 destination address 583 and port number. 585 +-------------------+ 586 | |----IPv6---\ 587 ----IPv4---\| |----IPv4---\\ 588 -----------/| |-----------// 589 | |-----------/ 590 | SMAP | 591 | | /--IPv6----- 592 /---IPv4----| |//---IPv4---- 593 \-----------| |\\----------- 594 | | \----------- 595 +-------------------+ 597 Figure 4: Stateless A+P Mapping Gateway Function 599 A SMAP-enabled node will perform the stateless 6/4 mapping function 600 for all public shared IPv4 addresses for which it was designated as a 601 stateless 6/4 mapping gateway. 603 To perform stateless 6/4 mapping function a SMAP gateway must: 605 o be provided with an IPv6 prefix (i.e., Pref6). The SMAP gateway 606 uses this prefix to construct IPv6 source addresses for all IPv4 607 shared addresses for which it was designated as a SMAP gateway. 608 The IPv6 prefix may be provisioned statically or dynamically 609 (e.g., DHCP) 611 o be able to know the IPv6 prefix of the node serving as another 612 SMAP gateway for IPv4 destination addresses. This prefix may be 613 known in various ways: 615 * Default or Well known prefix which was provisioned statically 616 or dynamically; 618 * Retained at the reception of incoming IPv4-in-IPv6 encapsulated 619 packets; 621 * Discovered at the communication starting thanks to mechanisms 622 as DNS resolution for example. 624 When the SMAP-enabled node receives IPv4 packets with IPv4 source 625 addresses for which it was not designated as a SMAP gateway, it will 626 not perform stateless 6/4 mapping function for those packets. Those 627 packets will be handled in a classical way (i.e., forwarded, dropped 628 or locally processed). 630 When the SMAP-enabled node receives IPv6 packets with IPv6 addresses 631 which do not match with its IPv6 prefix, it will not perform the 632 stateless 6/4 mapping function for those packets. Those packets will 633 be handled in a classical way (i.e., forwarded, dropped or locally 634 processed). 636 5.2. Implementation modes 638 Stateless mapping function may be achieved in two main modes. Those 639 modes consist in mapping the traffic only in one direction or in the 640 two directions as described below. 642 5.2.1. SMAP to route incoming traffic destined to a shared IPv4 address 644 IPv4 traffic with shared IPv4 source addresses are forwarded by the 645 node A without performing stateless mapping function. This traffic 646 will reach its destination thanks to a classical routing. In the 647 opposite direction, the traffic sent by the destination has to pass 648 by the node B which performs the stateless mapping function 649 (encapsulating in IPv6 packets) before forwarding to the node A. The 650 node A performs the stateless mapping function (extract IP v4 651 packets) before forwarding IPv4 packets to the points identified by 652 the IPv4 destination addresses and port number. In this case, both 653 IPv4 and IPv6 traffic are routed in the network between the nodes A 654 and B. 656 +----------+ 657 --IPv4---|----------|------------IPv4--------------------\ 658 ---------|----------|------------------------------------/ 659 | | 660 | +------+ | +------+ 661 | | | | /----IPv6-----| | 662 /--IPv4--| | SMAP | |//---IPv4---- | SMAP |/---IPv4---- 663 \--------| | | |\\----------- | |\----------- 664 | | | | \-------------| | 665 | +------+ | +------+ 666 | | 667 +----------+ 668 node A node B 670 Figure 5: First Configuration 672 5.2.2. No IPv4 capabilities are used anymore between two SMAP-enabled 673 nodes 675 In this configuration, the node A performs the stateless mapping 676 function on the received IPv4 traffic (encapsulated in IPv6 packets) 677 before forwarding to the node B. The node B performs the stateless 678 mapping function on the received IPv6 traffics (extracting IPv4 679 packets) before forwarding the IPv4 traffic to the destination 680 identified by the IPv4 destination address and port number. In the 681 opposite direction and as previously, the node B performs the 682 stateless mapping function on the received IPv4 traffics 683 (encapsulating in IPv6 packets) before forwarding to the node A. The 684 node A performs the stateless mapping function on the received IPv6 685 traffic (extracting IPv4 packets) before forwarding the IPv4 traffic 686 to the point identified by the IPv4 destination address and port 687 number. In this case, only IPv6 traffic is managed in the network 688 segment between the nodes A and B. 690 +------+ +------+ 691 | |----IPv6---\ | | 692 ----IPv4---\| |----IPv4---\\| |----IPv4---\ 693 -----------/| |-----------//| |-----------/ 694 | |-----------/ | | 695 | SMAP | | SMAP | 696 | | /----IPv6---| | 697 /---IPv4----| |//---IPv4----| |/---IPv4---- 698 \-----------| |\\-----------| |\----------- 699 | | \-----------| | 700 +------+ +------+ 701 node A node B 703 Figure 6: Second Configuration 705 5.3. Deployment Scenarios 707 Several deployment scenarios of the SMAP function may be envisaged in 708 the context of Port Range based solutions: 710 o A SMAP function is embedded in a port-restricted device. Other 711 SMAP-enabled nodes are deployed in the boundaries between IPv6- 712 enabled realms and IPv4 ones. This scenario may be particularly 713 deployed for intra-domain communications so as to interconnect 714 heterogeneous realms (i.e., IPv6/IPv4) within the same AS. 716 o A SMAP function is embedded in a port-restricted device. Other 717 SMAP-enabled nodes are deployed in the interconnection segment 718 (with adjacent IPv4-only ones) of a given AS. This deployment 719 scenario is more suitable for service providers targeting the 720 deployment of IPv6 since it eases the migration to full IPv6. 721 Core nodes are not required to activate anymore both IPv4 and IPv6 722 transfer capabilities. 724 Other considerations regarding the interconnection of SMAP-enabled 725 domains should be elaborated. The following provides a non 726 exhaustive list of interconnection schemes. 728 o The interconnection of two domains implementing the SMAP function 729 may be deployed via IPv4 Internet (Figure 7): This means that IPv4 730 packets encapsulated in IPv6 one are transferred using IPv6 until 731 reaching the first SMAP-node. Then these packets are de- 732 encapsulated and are forwarded using IPv4 transfer capabilities. 733 A remote SMAP-enabled node will receive those packets and proceeds 734 to an IPv4-in-IPv6 encapsulation. These packets are then routed 735 normally until reaching the port-restricted devices which de- 736 encapsulates the packets. 738 +------+ +------+ +--------+ +------+ +------+ 739 | |--IPv6---\ | | | | | |---IPv6--\ | | 740 | |--IPv4---\\| |---|-IPv4---|--\| |---IPv4--\\| | 741 | |---------//| |---|--------|--/| |---------//| | 742 | |---------/ | | |Internet| | |---------/ | | 743 | SMAP | | SMAP | | IPv4 | | SMAP | | SMAP | 744 | | /---IPv6--| | | | | | /---IPv6--| | 745 | |//---IPv4--| |/--|-IPv4---|---| |//--IPv4---| | 746 | |\\---------| |\--|--------|---| |\\---------| | 747 | | \---------| | | | | | \---------| | 748 +------+ +------+ +--------+ +------+ +------+ 749 Source node A node B Destination 751 Figure 7: Interconnection Scenario 1 753 o A second scheme is to interconnect two realms implementing the 754 SMAP function using IPv6 (Figure 8). Two sub-scenarios are 755 identified: 757 * An IPv6 prefix (i.e., Pref6) assigned by IANA is used for this 758 service. If appropriate routing configuration have been 759 enforced, then the IPv6 encapsulated packets will be routed 760 until the final destination. 762 * If an IPv6 belonging a service provider prefix is used. This 763 will be covered in the next versions of the document. 765 +------+ +------------+ +------+ 766 | | | | | | 767 | |----IPv6-----|----IPv6----|----IPv6----\ | | 768 | |----IPv4-----|------------|----IPv4----\\| | 769 | |-------------|------------|------------//| | 770 | |-------------|------------|------------/ | | 771 | SMAP | | Internet v6| | SMAP | 772 | | /-----IPv6--|------------|-----IPv6-----| | 773 | |//---IPv4----|------------|-------IPv4---| | 774 | |\\-----------|------------|--------------| | 775 | | \-----------|------------|--------------| | 776 | | | | | | 777 +------+ +------------+ +------+ 778 Source Destination 780 Figure 8: Interconnection Scenario 2 782 5.4. SMAP and PRR 784 Within this draft, a PRR-enabled node implements both SMAP function 785 and required functions to handle fragmentation and other protocols 786 which do not require a transport protocol. More details about 787 fragmentation may be found at [I-D.boucadair-port-range]. 789 In the remaining part, the text refers only to PRR and not SMAP. 791 6. IPv6 Migration Scenarios 793 6.1. Overview 795 This section proposes a set of migration steps in the context of IPv4 796 address exhaustion and IPv6 deployment. Both objectives are taken 797 into account. 799 The proposed steps are informational. An analysis of these steps and 800 proposed IPv6 migration paths are discussed in Section 6.7. 802 The following figure (i.e., (Figure 9)) provides an overview of 803 network segments and the localisation of PRR-enabled nodes. One or 804 several PRR may be enabled. PRD1 and PRD2 are two port-restricted 805 devices which have been provisioned with the same IP1 public address 806 and two distinct port ranges (PR1 and PR2). 808 +---------+ +---------------------+ +----------+ +---------+ 809 +----+ | | | | | +-----+ | |IPv4 | 810 |PRD1|----| |--| |--| |i-PRR| |--|Internet | 811 +----+ | +-----+ | | | | | | | +---------+ 812 IP1, PR1 | |a-PRR| | | core network | | +-----+ | 813 | +-----+ | | | | | 814 +----+ | | | | | | +---------+ 815 |PRD2|----| | | | | |--|IPv6 | 816 +----+ | | | | | | |Internet | 817 IP1, PR2 +---------+ +---------------------+ +----------+ +---------+ 818 access/ interconnection 819 backhaul segment 821 Figure 9: Reference Architecture 823 6.2. IPv6 Prefixes and Addresses 825 Different types of IPv6 prefixes and addresses are used in the scope 826 of the solutions described in the document (i.e., Step_0 827 (Section 6.4), Step_1 (Section 6.5) and Step_2 (Section 6.6)). 828 Theses prefixes and addresses are listed hereafter: 830 1. IPv6Pref: A prefix allocated to the port-restricted device. A 831 packet sent to addresses belonging to this prefix are routed 832 toward this port-restricted device. IPv6Pref prefix addresses 833 may also be used to send and receive native IPv6 traffic. In 834 stateless IPv6-IPv4 Address Mapping mode (as explained above), 835 the IPv6Pref structure is related to the IPv4 address plus port 836 range. In binding mode, IPv6Pref and IPv4 address plus port 837 range are independent. 839 2. IPv6PrefA: An address belonging to IPv6Pref prefix used to send 840 IPv4-in-IPv6 traffic. 842 3. Pref6: An IPv6 prefix (e.g., /24, /32) common to all of the IPv6 843 packets which must be routed to a PRR function. It is for 844 further study to decide whether this prefix is to be: 846 A. Service provider scope 848 B. or common to all service providers (to be defined by IANA). 850 Both alternatives are compatible with the proposed solutions. 852 4. Pref6A: An address belonging to the Pref6 prefix. A Pref6A 853 address includes the Pref6 prefix on its left most part followed 854 by the destination IPv4 address and destination port number, as 855 shown in Figure 3. When a binding table is implemented, a given 856 a-PRR has to transform a destination address Pref6A to a 857 destination address IPv6PrefA, it proceeds as illustrated in 858 Figure 10. 860 +-------------------------------------+-------------------------------------+ 861 | | | | | 862 | Pref6 |public IPv4 |Dest. | 0:0:0:0 | 863 | |address |port | | 864 +---------------------------------------------------------------------------+ 865 | || | 866 || 867 vv 868 +--------------------------------------------------+ 869 | binding table | 870 | | 871 |(...) | 872 | [public IPv4 address, Port Range] => IPv6PrefA | 873 |(...) || | 874 +---------------------------------------------||---+ 875 || 876 /====================/ 877 || 878 \/ 879 +---------------------------------------------------------------------------+ 880 | | 881 | IPv6PrefA | 882 | | 883 +---------------------------------------------------------------------------+ 885 Figure 10: Fetching IPv6PrefA from WKIPv6A 887 The following sections describe three migration steps and a set of 888 proposed migration paths. The proposed solutions are stateless at 889 interconnection segment. A binding table may be implemented to meet 890 requirements of service providers which do not want to closely 891 correlate their IPv6 address plan and the IPv4 one. More details are 892 provided below. 894 6.3. On Stateless and Binding Table Modes 896 6.3.1. Stateless Mode 898 Complete stateless mapping implies that the IPv4 address and the 899 significant bits coding the port range are reflected inside the IPv6 900 prefix assigned to the port-restricted device. Two alternatives are 901 offered when such a stateless mapping is to be enabled: 903 - either using the IPv6 prefix already used for native IPv6 904 traffic, 906 - or provide two prefixes to the port-restricted device: one for 907 the native IPv6 traffic and one for the IPv4 traffic. 909 Note that: 911 - Providing two IPv6 prefixes has the advantages of allowing a /64 912 prefix for the port-restricted device along with another prefix 913 (e.g., a /56 or /64) for native IPv6 traffic. This alternative 914 spares the service provider to relate the native IPv6 traffic 915 addressing plan to the IPv4 addressing plan. The drawback is the 916 burden to allocate two prefixes to each port-restricted device and 917 to route them. In addition, an address selection issue may be 918 encountered. 920 - Providing one prefix for both needs (e.g., a /56 or a /64) 921 spares the service provider to handle two types of IPv6 prefix for 922 the port-restricted device and in routing tables. But the 923 drawback is that it somewhat links strongly the IPv4 addressing 924 plan to the allocated IPv6 prefixes. 926 6.3.2. Binding Table Mode 928 Another alternative is to assign a "normal" IPv6 prefix to the port- 929 restricted device and to use a binding table, which can be hosted by 930 a service node, to correlate the (shared IPv4 address, Port Range) 931 with an IPv6 address part of the assigned IPv6 prefix. For 932 scalability reasons, this table should be instantiated within PRR- 933 enabled nodes which are close to the port-restricted devices. The 934 number of required entries if hosted at interconnection segment would 935 be equal to the amount of subscribed users (one per port-restricted 936 device). 938 The stateless mode is recommended. 940 6.4. Step_0: IPv6 at Access Network 942 This step is described in [I-D.boucadair-port-range]. This step 943 assumes that IPv6 is used to convey incoming traffic to its final 944 destination. For this reason, an IPv6 address is used as the routing 945 identifier. More information about this step is provided below. 947 6.4.1. Context 949 This step can be deployed at earlier stages of IPv6 deployment. The 950 impact on routing (especially path optimisation) and also offered 951 services is the same as for the Port Range solution described in 952 [I-D.boucadair-port-range]. The service brokenness risk is 953 optimized. IPv6 is used in this step as a means to convey incoming 954 IPv4 traffic. Within this step, IPv6 is used in the access segment 955 to deliver the received IPv4 traffic. 957 When this step is deployed, at least 50% of the handled traffic 958 (incoming+outgoing), at the IP access segment, of a service 959 provider's domain is achieved using IPv6 capabilities. IPv4 960 capabilities are used only for outgoing traffic. 962 Native IPv6 connectivity may also be offered to end-users. 964 6.4.2. Overall Procedure 966 This section discusses additional points related to IPv6 usage in the 967 context of the Port Range solution described in 968 [I-D.boucadair-port-range]. 970 6.4.2.1. Provisioning Operations 972 This section lists the set of information required for a port- 973 restricted device to access the connectivity service. 975 6.4.2.1.1. IP Connectivity Information 977 Each port-restricted device is assigned with: 979 1. A shared public IPv4 address. In addition to this address, a 980 port range is also assigned to the device. 982 2. An IPv6 prefix: denominated in the rest as IPv6Pref. This prefix 983 is allocated to the port-restricted device and it is used to 984 discriminate (a given device) among all those having the same 985 public IPv4 address. This address may be used also to send and 986 receive native IPv6 traffic or a second prefix may be assigned 987 specific for native IPv6 traffic. 989 6.4.2.1.2. Provisioning procedure 991 To convey IPv4 configuration information, one of these solutions may 992 be implemented: 994 1. Activate DHCP and support port range options as described in 995 [I-D.bajko-pripaddrassign]; 997 2. Use PPP and support the IPCP Port Range Configuration Options as 998 specified in [I-D.boucadair-pppext-portrange-option]. 1000 To convey IPv6 configuration information, DHCPv6 [RFC3315] may be 1001 activated. When several IPv4-inferred IPv6 address structures are 1002 supported, options defined in 1003 [I-D.boucadair-dhcpv6-shared-address-option] can be used. 1005 6.4.2.2. Port Restricted Device's behaviour and supported functions 1007 A port-restricted device may be a host, a CPE, etc. 1009 6.4.2.2.1. Port Restriction Behaviour 1011 The behaviour of the port-restricted device is as follows: 1013 1. If the port-restricted device hosts a NAT function: For incoming 1014 traffic, the port-restricted device checks if the destination 1015 port number is within the Port Range, otherwise the packet is 1016 dropped. When the destination port number of a received packet 1017 (from outside the LAN) falls inside the Port Range, classical NAT 1018 operations are enforced and the packet is then routed to its 1019 final destination in the LAN. 1021 2. Otherwise, the port-restricted device is an end-user host: the 1022 device restricts its source port numbers to be with the assigned 1023 Port Range. Received IPv4 packets with a destination port number 1024 outside the Port Range must be dropped. 1026 6.4.2.2.2. Handling Outgoing traffic 1028 The same procedure as described in [I-D.boucadair-port-range] 1029 applies. In addition to the normal NAT operations, the port- 1030 restricted device ensures that the source port number is within the 1031 allowed Port Range. 1033 6.4.2.2.3. Handling Incoming traffic 1035 For incoming traffic, two cases may be considered: 1037 1. IPv6 encapsulated traffic: for encapsulated IPv6 packets, the 1038 port-restricted device de-encapsulates the packets and extracts 1039 the embedded IPv4 one. The original IPv4 packets is then treated 1040 and handled locally. If the destination port of that packet is 1041 within the Port Range of that port-restricted device, and 1042 depending on the local NAT implementation if any, the packet may 1043 be accepted and then proceed to classical NAT operation. 1044 Otherwise, the packet is dropped. 1046 2. IPv6 native traffic: No constraint is required. The traffic 1047 should be routed to it final destination, if the port-restricted 1048 device is a CPE. 1050 6.4.2.3. PRR Behaviour 1052 6.4.2.3.1. Supported functions 1054 In addition to the functions listed in [I-D.boucadair-port-range], 1055 the PRR must support an IPv6 encapsulation function. 1057 6.4.2.3.2. Localization 1059 The PRR function is deployed under the same conditions as the ones 1060 discussed in [I-D.boucadair-port-range]. 1062 6.4.2.3.3. Behaviour 1064 The PRR intervenes only for incoming traffic destined to a shared 1065 IPv4 address. 1067 a. If a binding table is implemented: This binding table stores the 1068 required information to route the traffic destined to a shared IP 1069 address to the appropriate port-restricted device among all those 1070 sharing the same IP address. An IPv6 prefix may be used as 1071 routing identifier. In this case, the structure of the binding 1072 table is: (shared IPv4 address, Port Range) ==> IPv6 prefix. 1073 Instead of the IPv6 prefix itself (IPv6Pref), the binding table 1074 may contain a specific address under IPv6Pref (called in the rest 1075 IPv6PrefA). When a binding table is adopted, the IPv6 prefix 1076 assigned to a port-restricted device is not constrained. There 1077 is no need for the service provider to allocate two different 1078 IPv6 prefixes to the port-restricted devices (one for native IPv6 1079 traffic and another one for the IPv4 encapsulated traffic). Only 1080 one may suffice for the two needs. 1082 b. Stateless mapping: The service provider assigns an IPv4 shared 1083 address, a port range and an IPv6 prefix with IPv6 prefix 1084 containing explicitly the IPv4 address and the significant bits 1085 of the port range (i.e., Bits used to built the Port Range Mask, 1086 see [I-D.bajko-pripaddrassign]). When a stateless mapping is 1087 adopted, it is possible that the service provider has to cope 1088 with constraints when allocating the IPv6 prefix and the shared 1089 IPv4 address to the port-restricted devices. As a matter of fact 1090 the IPv6 prefix must reflect the shared IPv4 address. 1091 Alternatively the service provider may instead allocate two 1092 different prefixes for the two needs (IPv6 native traffic and 1093 IPv4 encapsulated traffic). 1095 In the remaining part of this section, "PRR retrieves the 1096 corresponding IPv6 prefix address" means that: 1098 a. If a binding table is implemented: the PRR looks-up through this 1099 table and retrieves the IPv6Pref or IPv6PrefA corresponding to 1100 (IPv4 address, Port Range). If IPv6Pref is retrieved, the PRR 1101 builds an IPv6Pref address in complementing the IPv6Pref with a 1102 fixed bit sequence chosen by the service provider to be always 1103 the same complementary bit sequence for all port-restricted 1104 devices. . 1106 b. Stateless mode: an IPv6 prefix address (i.e., IPv6PrefA) is built 1107 using the IPv4 address and the destination port number. 1109 In both cases, when the PRR has got/built the corresponding IPv6PrefA 1110 , the PRR encapsulates the original packets in an IPv6 one with a 1111 destination IP address equal to the IPv6Pref address. The source 1112 IPv6 address is an address of the PRR. It may be an anycast IPv6 1113 address or unicast one. 1115 This packet is then routed according to instantiated IGP routes. 1117 6.4.2.4. Routing considerations 1119 The same IGP considerations as detailed in [I-D.boucadair-port-range] 1120 should be taken into account. In addition to these considerations, 1121 IPv6 routes should be installed to reach port-restricted devices from 1122 an IPv6-enabled PRR. 1124 6.4.3. Focus on Communication Establishment 1126 6.4.3.1. Outgoing IPv4 Communications 1128 Outgoing IPv4 traffic is handled as described in 1129 [I-D.boucadair-port-range]. The traffic issued from a port- 1130 restricted device is routed to its final destination. The traffic is 1131 not altered and is transferred to its final destination. 1133 6.4.3.2. Incoming IPv4 Communications 1135 Owing to IGP configuration, the traffic destined to a shared IP 1136 address must cross a PRR. This latter encapsulates the received IPv4 1137 packets in IPv6 ones as described in Section 6.4.2.3.3. The traffic 1138 is then routed using the IPv6PrefA as a destination address. The 1139 traffic is received by the device among those sharing the same IPv4 1140 address (because this port-restricted device is allocated with this 1141 IPv6Pref). A de-encapsulation operation is then executed as 1142 described in Section 6.4.2.2. If the de-encapsulated IPv4 traffic is 1143 destined to a port within the assigned Port Range, the traffic is 1144 accepted, otherwise it is dropped. 1146 6.4.3.3. Outgoing IPv6 Communications 1148 Since the port-restricted device is IPv6-enabled, native IPv6 1149 communications may be offered. This assumes that the service 1150 provider has deployed means for IPv6 transfer capability. The same 1151 prefix used to convey incoming IPv4 traffic (IPv6Pref) may be used 1152 also to send and receive native IPv6 traffic. Alternatively, a 1153 second IPv6 prefix may be assigned to that purpose. 1155 6.4.3.4. Incoming IPv6 Communications 1157 Native IPv6 communications are supported. 1159 6.4.4. Typical Flow Example 1161 In order to illustrate this step, let consider the example shown in 1162 the following figure (Figure 11). M1 is a machine behind a port- 1163 restricted device (called CPE as Customer Port restricted Equipment 1164 in the example and Figure 11). 1166 M1 wants to establish an IPv4 communication with RM (Remote Machine). 1167 To do so, an IPv4 packet is issued by M1. This packet has as source 1168 IP address equal to Pri_IPv4. The packet is then received by CPE. 1169 This latter enforces its NAT operations. As a result, an IPv4 packet 1170 with a source IP address equal to Pub_IPv4 and a source port number 1171 within the Port Range of CPE is sent. The resulting packet is 1172 forwarded according to IPv4 transfer capabilities until reaching its 1173 final destination RM. 1175 As a response, RM sends an IPv4 packet destined to Pub_IPv4 and a 1176 destination port number equal to the source port number of the 1177 received packet. This message is received by PRR. The PRR 1178 encapsulates the received IPv4 packet in an IPv6 one. The resulting 1179 IPv6 packet is then forwarded. The encapsulated packet is received 1180 by the appropriate CPE among those having the same IPv4 address. A 1181 de-encapsulation is enforced. The original IPv4 packet is extracted. 1182 Once classical NAT operations are executed, the CPE forwards the IPv4 1183 packet to M1. 1185 +--+ +---+ +--+ 1186 |M1| |CPE| |RM| 1187 +--+ +---+ +--+ 1188 | | | 1189 |==Pri_IPv4_Out==>|===============Pub_IPv4_Out===============>| 1190 | | | 1191 | | +---+ | 1192 | | |PRR| | 1193 | | +---+ | 1194 | | | | 1195 |<==Pri_IPv4_In===|<==IPv6_Enc(Pub_IPv4_In)==|<==Pub_IPv4_In==| 1196 | | | | 1198 Figure 11: Flow Example (Step 0): Inter-domain 1200 6.5. Step_1: IPv6 a Means to Transfer Incoming IPv4 Packets 1202 6.5.1. Context 1204 Step_1 is characterized by the activation of two levels of PRR 1205 functions in several segments of a given service provider's network. 1206 Some PRR-enabled nodes are deployed close to the port-restricted 1207 devices (e.g., In the access or backhaul network) whilst others are 1208 installed at the interconnection segment of the service provider's 1209 network as shown in Figure 9. 1211 The objective of this step is to maximize the invocation of IPv6 1212 capabilities, particularly to convey incoming IPv4 traffic until 1213 delivery to final destination (e.g., port-restricted device). 1215 6.5.2. Overall Procedure 1217 Step_1 works exactly the same way as that the Step_0, apart what is 1218 specified in this section. In particular, there are none difference 1219 between Step_0 and Step_1 at the level of the port-restricted device. 1221 6.5.2.1. Routing considerations 1223 - a-PRR: As in Step_0, a-PRR announces the shared IPv4 prefixes it 1224 serves. In addition, in case of binding table mode, it announces 1225 in IGP the aggregates of all the Pref6A addresses of the port- 1226 restricted devices it serves so that IPv4-in-IPv6 packets reach 1227 this a-PRR. 1229 - i-PRR: i-PRR announces in EGP (if embedded in an ASBR node) or 1230 in IGP (if deployed behind an ASBR), all (aggregated) IPv4 1231 prefixes of all port-restricted devices it can route packets to 1232 (via a-PRR in case of binding table). Depending of the structure 1233 of the service provider network, some i-PRR-enabled nodes may be 1234 positioned inside the service provider network to encapsulate more 1235 IPv4 traffic into IPv6. For example, from a region A to another 1236 region B of the service provider's network. 1238 6.5.2.2. Behaviour of a-PPR and i-PRR 1240 Figure 12 show the respective role of a-PRR and i-PRR-enabled nodes. 1241 The labels of the arrows are explained in further sub-sections. 1243 - CPE1 (Customer Port restricted Equipment) is a port-restricted 1244 device 'served' by a-PRR (means that a-PRR announces in IGP the 1245 assigned shared IPv4 address to CPE1). 1247 - CPE2 is a device of another customer connected to the network 1248 managed by the same service provider. It may be either another 1249 port-restricted device or a device with a plain IPv4 address. 1251 - RM is a remote machine located outside the AS managed by the 1252 service provider. 1254 +----+ +-----+ +-----+ 1255 |CPE1| |i-PRR| | RM | 1256 +----+ +-----+ +-----+ 1257 | | | 1258 |<=====IPv6PrefA_Enc(Pub_IPv4_In)=======|<===Pub_IPv4_In===| 1259 | | | 1260 | | | 1261 | +-----+ 1262 | |a-PRR| 1263 | +-----+ +----+ 1264 | | |CPE2| 1265 | | +----+ 1266 | | | 1267 |<==IPv6PrefA_Enc |<==Pub_IPv4_In==| 1268 | (Pub_IPv4_In)==| | 1269 | | | 1271 Figure 12: Step_1: Stateless mode 1273 +----+ +-----+ +-----+ +-----+ 1274 |CPE1| |a-PRR| |i-PRR| | RM | 1275 +----+ +-----+ +-----+ +-----+ 1276 | | | | 1277 |<==IPv6PrefA_Enc |<==Pref6A_Enc |<===Pub_IPv4_In===| 1278 | (Pub_IPv4_In)==| (Pub_IPv4_In)=| | 1279 | | | | 1280 | | 1281 | | 1282 | | +----+ 1283 | | |CPE2| 1284 | | +----+ 1285 | | | 1286 |<==IPv6PrefA_Enc |<==Pub_IPv4_In==| 1287 | (Pub_IPv4_In)==| | 1288 | | | 1290 Figure 13: Step_1: Binding table mode 1292 6.5.2.2.1. Localization 1294 The PRR function is deployed under the same conditions as in Step_0 1295 but as previously mentioned more PRR-enabled nodes are deployed 1296 within the service provider's network. 1298 6.5.2.2.2. Stateless Mapping Mode 1300 In case the service provider assigns an IPv4 shared address, a port 1301 range and an IPv6 prefix containing explicitly the IPv4 address and 1302 the significant bits of the port range, i-PRR-nodes are able to build 1303 the IPv6Pref address of the port-restricted device using the IPv4 1304 destination address and destination port bits of received IPv4 1305 packets. 1307 Hence, i-PPR behaviour is the same one as the PRR one described in 1308 Step_0, when stateless mapping is enforced. In that case, the IPv4- 1309 in-IPv6 packet does not pass through the a-PRR but it is routed 1310 directly to the port-restricted device (e.g., To CPE1 as depicted in 1311 Figure 12). 1313 6.5.2.2.3. Binding Table Mode 1315 This behaviour is illustrated in Figure 13. 1317 If a binding table is implemented within a-PRR, i-PRR-enabled nodes 1318 can not hold all the binding table entries corresponding to all the 1319 port-restricted devices it may route traffic for. Consequently, it 1320 has to route the IPv4 traffic towards the a-PRR of which the port- 1321 restricted device depends. More precisely, a given i-PRR 1322 encapsulates the incoming IPv4 traffic in IPv6 packets using the 1323 following addresses: 1325 - The source IPv6 address is one of the global IPv6 addresses of 1326 the i-PRR. 1328 - The destination IPv6 is built by i-PRR using an address under 1329 the Pref6 prefix conforming to the formalism defined in 1330 Section 4.2. This address, called Pref6A, includes the Pref6 1331 prefix on its left most part and the destination IPv4 address and 1332 destination port number in this right most part. 1334 Thus, an i-PRR-enabled node routes normally this IPv4-in-IPv6 packet 1335 (labeled Pref6A_Enc (Pub_IPv4_In) in Figure 13) using IPv6 transfer 1336 capabilities. The packet is routed towards the a-PRR serving the 1337 recipient port-restricted device (CPE1 is Figure 13) since 1338 appropriate routing configuration has been enforced (see 1339 Section 6.5.2.1). 1341 Upon receipt of that packet, the a-PRR proceeds as follows: 1343 - It retrieves the IPv4 shared address and port bits parts of the 1344 Pref6A address. Then, with these parts it looks-up through its 1345 binding table to fetch the IPv6PrefA corresponding to the couple 1346 (IPv4 address, Port Range). 1348 - Once retrieved, the a-PRR positions the IPv6PrefA address as the 1349 destination address in place of the Pref6A address and forwards 1350 the packet. The IPv4-in-IPv6 packet is routed until reaching the 1351 port-restricted device (CPE1 in Figure 13). 1353 As shown in Figure 13 (bottom part), when another machine (CPE2) 1354 within the same service provider's domain sends traffic to the port- 1355 restricted device CPE1, the working of a-PRR is the same one as that 1356 of the PRR is Step_0. 1358 6.6. Step_2: Only IPv6 Is Used For Both Incoming and Outgoing Traffic 1360 6.6.1. Context 1362 This step is suitable for service providers wishing to migrate to 1363 full IPv6 and to offer a global connectivity using IPv6. This step 1364 provides a lightweight procedure to interconnect IPv6 with IPv4 1365 realms. This procedure may be fully stateless or require a binding 1366 table. This table does not include session-based information. 1368 Only IPv6 connectivity is used inside the service provider's domain; 1369 IPv4 capabilities are deactivated. No parallel IPv4 and IPv6 1370 operational tasks will be maintained anymore in the core segment. 1372 Two implementation modes may be envisaged: 1374 1. Encapsulation-based mode: This mode suggests using both inbound 1375 and outbound IPv6 encapsulation to carry IPv4 traffic (received 1376 from the remaining IPv4-only realms). This mode is almost 1377 similar to Step_1 for handling incoming IPv4 packets. Unlike 1378 Step_1, the required operations to build the outgoing 1379 encapsulated packets is also supported in this step. 1381 2. Translation-based mode: Unlike the first mode, this one assumes 1382 that there is no need to maintain both IPv4 and IPv6 stacks in 1383 the CPE. It is recommended to implement this mode when mature 1384 IPv6 deployment has been observed and that IPv6 realms become 1385 more important than IPv4-only ones. 1387 More information about these two modes is provided in the sub- 1388 sections hereafter. 1390 6.6.2. The IPv6 Encapsulation-Based Mode 1392 6.6.2.1. Provisioning Operations 1394 6.6.2.1.1. IP Connectivity Information 1396 In addition to the information required for Step_1, a Pref6 IPv6 1397 prefix may e configured on the port-restricted device. This prefix 1398 is to be used when running the IPv4-to-IPv6 mapping function required 1399 to encapsulate IPv4 traffic in IPv6 one. 1401 6.6.2.1.2. Provisioning Procedure 1403 Idem as Step_1. 1405 6.6.2.2. Routing Considerations 1407 IPv4 IGP protocols are not anymore enabled in the core network. Only 1408 IPv6 routing table is maintained by involved routers. 1410 Inter-domain IPv4 connectivity is maintained with IPv4-only realms. 1411 IPv4 network prefixes are mapped to IPv6 prefixes (using a Pref6 1412 configured by the service provider) which are injected in the 1413 deployed IPv6 IGP protocol. 1415 A given a-PRR MUST advertise in IGP the aggregated IPv6 prefixes it 1416 handles. Doing so, all intra-domain IPv6 packets will cross that 1417 PRR. 1419 6.6.2.3. Port Restricted Device's Behaviour and Supported Functions 1421 6.6.2.3.1. Port Range Restriction 1423 Idem as Step_1. 1425 6.6.2.3.2. Handling Outgoing Traffic 1427 Unlike Step_1, outgoing IPv4 traffic is encapsulated in IPv4-in-IPv6 1428 packets. Concretely, the port-restricted device executes its port 1429 restricted NAT operations (if any). The resulting IPv4 packet is 1430 then encapsulated in an IPv6 packet. The port-restricted device 1431 selects an IPv6 address from its assigned IPv6 prefix (IPv6Pref). 1432 This address IPv6PrefA is used as the source IPv6 address of the 1433 encapsulated packet. 1435 Two options may be considered to build the destination IPv6 address 1436 of the encapsulated packet as listed below: 1438 1. The port-restricted device is provisioned to use an anycast IPv6 1439 address. This anycast IPv6 address is configured on internal 1440 interfaces of all PRRs. This mode is may be implemented when the 1441 port-restricted device is not able to build a destination IPv6 1442 address reflecting the IPv4 address and port of its correspondent 1443 (i.e., the port-restricted device does not support the mapping 1444 function defined in Section 4.2). 1446 2. The port-restricted device is able to build an IPv6 address using 1447 a Pref6 prefix (which may be distinct than the one used to build 1448 mapped IPv6 prefixes by i-PRRs), the destination IPv4 address and 1449 the destination port number. 1451 No constraints are to be followed for outgoing native IPv6 traffic. 1453 6.6.2.3.3. Handling Incoming Traffic 1455 Idem as Step_1. 1457 6.6.2.4. PRR Behaviour 1459 6.6.2.4.1. Supported Functions 1461 In addition to the functions supported in Step_1, an IPv4-in-IPv6 de- 1462 encapsulation function must be supported by a-PRR. 1464 6.6.2.4.2. Localization 1466 Idem as Step_1. 1468 6.6.2.4.3. Behaviour: Stateless Mode 1470 The behaviour of both access and interconnection PRRs is elaborated 1471 below: 1473 1. If an anycast IPv6 address is configured on interfaces of all 1474 a-PRRs: 1476 * An (access) PRR will receive the encapsulated packet issued 1477 from port-restricted devices. The packet is de-encapsulated 1478 and the original IPv4 one is retrieved. Then, the (access) 1479 PRR builds a destination IPv6 address using a Pref6 prefix, 1480 the destination IPv4 address and the destination port number. 1481 The original IPv4 packet is then encapsulated in IPv6 packet 1482 with a source IPv6 address of the (access) PRR and the 1483 destination IPv6 address equal to the newly built one. The 1484 packet is forwarded to next hop according to IPv6 routing 1485 table. If the correspondent is not a port-restricted device, 1486 the packet is intercepted by a-PRR or a i-PRR depending of 1487 where the correspondent is located. This a-PRR or i-PRR 1488 proceeds to the de-encapsulation operation. The extracted 1489 IPv4 packet is then forwarded to the IPv4 correspondent. This 1490 encapsulated packet is received by an (access/ 1491 interconnection) PRR which proceeds to a de-encapsulation 1492 operation. The extracted IPv4 packet is then forwarded to the 1493 next IPv4 hop. 1495 * Incoming IPv4 traffic is intercepted by an (interconnection) 1496 PRR. The PRR encapsulates the received IPv4 packet in an IPv6 1497 one using the following information: 1499 + The source IPv6 address is one of the global IPv6 addresses 1500 of the PRR. 1502 + The destination IPv6 is built by the PRR using the 1503 formalism defined in Section 4.2. This address is build 1504 using a Pref6 prefix, the destination IPv4 address and the 1505 destination port number. The appropriate port-restricted 1506 device, among those having the same IPv4 address, will 1507 receive the packet. This is illustrated in Figure 14. 1509 +---+ +-----+ +-----+ +--+ 1510 |CPE| |a-PRR| |i-PRR| |RM| 1511 +---+ +-----+ +-----+ +--+ 1512 | | | | 1513 |=IPv6_Enc(Pub_IPv4_Out)==>|=Pref6A_Enc(Pub_IPv4_Out)==>|==Pub_IPv4_Out=>| 1514 | | | | 1515 | | | | 1516 |<=========IPv6PrefA_Enc(Pub_IPv4_In)===================|<==Pub_IPv4_In==| 1517 | | | | 1518 | | | | 1520 LAN messages are not represented in the figure. 1522 Figure 14: Encapsulation Mode: Anycast addresses are 1523 assigned to all PRR 1525 2. Otherwise, all internal IPv4 traffic is encapsulated by the port 1526 restricted device in IPv4-in-IPv6 packets (the destination IPv6 1527 address is the one built by the port-restricted device, see 1528 Section 4.2). As in the first bullet, the packet is forwarded to 1529 next hop according to IPv6 routing table. If the correspondent 1530 is not a port-restricted device, the packet is intercepted by 1531 a-PRR or a i-PRR depending of where the correspondent is located. 1532 This a-PRR or i-PRR proceeds to the de-encapsulation operation. 1533 The extracted IPv4 packet is then forwarded to the IPv4 1534 correspondent. The same behaviour as for the first bullet 1535 applies for incoming IPv4 traffic. A flow is illustrated in 1536 Figure 15. 1538 +---+ +-----+ +--+ 1539 |CPE| |i-PRR| |RM| 1540 +---+ +-----+ +--+ 1541 | | | 1542 |=======Pref6A_Enc(Pub_IPv4_Out)================>|====Pub_IPv4_Out=>| 1543 | | | 1544 | | | 1545 |<======IPv6PrefA_Enc(Pub_IPv4_In)===============|<==Pub_IPv4_In====| 1546 | | | 1548 Figure 15: Encapsulation Mode: the port restricted device builds 1549 a destination IPv6 address 1551 6.6.2.4.4. Behaviour: Binding Table Mode 1553 The behaviour of both access and interconnection PRRs is elaborated 1554 below: 1556 1. If an anycast IPv6 address is configured on interfaces of all 1557 a-PRRs: 1559 * An (access) PRR will receive the encapsulated packets issued 1560 from port-restricted devices. The packets are de-encapsulated 1561 and the original IPv4 is retrieved. Then, the (access) PRR 1562 builds a destination IPv6 address using a Pref6 prefix and the 1563 destination IPv4 address. The original IPv4 packets are then 1564 encapsulated in IPv6 packet with a source IPv6 address of the 1565 (access) PRR and the destination IPv6 address equal to the 1566 newly built one. The packet is forwarded to next hop 1567 according to IPv6 routing table. The packet is intercepted by 1568 a-PRR or a i-PRR depending of where the correspondent is 1569 located. This a-PRR or i-PRR proceeds to the de-encapsulation 1570 operation. The extracted IPv4 packet is then forwarded to the 1571 IPv4 correspondent. 1573 * Incoming IPv4 traffic is intercepted by an (interconnection) 1574 PRR. The PRR encapsulates the received IPv4 packet in an IPv6 1575 one using the following information: 1577 + The source IPv6 address is one of the global IPv6 addresses 1578 of the PRR. 1580 + The destination IPv6 is built by the PRR using the 1581 formalism defined in Section 4.2. This address is build 1582 using a Pref6 prefix, the destination IPv4 address and the 1583 destination port number. The appropriate a-PRR managing 1584 the destination port-restricted device, among those having 1585 the same IPv4 address, will receive the packet. This is 1586 illustrated in Figure 16. The PRR de-encapsulates the 1587 packet and retrieves the original IPv4 packet. The access 1588 PRR retrieves the destination address IPv6PrefA stored in 1589 its binding table and forwards the packet to the port 1590 restricted device. 1592 +---+ +-----+ +-----+ +--+ 1593 |CPE| |a-PRR| |i-PRR| |RM| 1594 +---+ +-----+ +-----+ +--+ 1595 | | | | 1596 |==IPv6_Enc(Pub_IPv4_Out)=>|=Pref6A_Enc(Pub_IPv4_Out)==>|==Pub_IPv4_Out=>| 1597 | | | | 1598 |<==IPv6_Enc(Pub_IPv4_In)==|<==Pref6A_Enc(Pub_IPv4_In)==|<==Pub_IPv4_In==| 1599 | | | | 1601 LAN messages are not represented in the figure. 1603 Figure 16: Encapsulation Mode with a Binding table 1604 (Anycast) 1606 2. Otherwise, all internal IPv4 traffic is encapsulated by the port 1607 restricted device in IPv4-in-IPv6 packets. As in the first 1608 bullet, the packet is forwarded to next hop according to its IPv6 1609 routing table. The packet is intercepted by a-PRR or a i-PRR 1610 depending of where the correspondent is located. This a-PRR or 1611 i-PRR proceeds to the de-encapsulation operation. The extracted 1612 IPv4 packet is then forwarded to the IPv4 correspondent. The 1613 same behaviour as for the first bullet applies for incoming IPv4 1614 traffic. A flow example is illustrated in Figure 17. 1616 +---+ +-----+ +--+ 1617 |CPE| |i-PRR| |RM| 1618 +---+ +-----+ +--+ 1619 | | | 1620 |================Pref6A_Enc(Pub_IPv4_Out)===============>|=Pub_IPv4_Out=>| 1621 | | | 1622 | +-----+ | | 1623 | |a-PRR| | | 1624 | +-----+ | | 1625 | | | | 1626 |<=IPv6PrefA_Enc(Pub_IPv4_In)|<=Pref6A_Enc(Pub_IPv4_In)==|<=Pub_IPv4_In==| 1627 | | | | 1629 Figure 17: Encapsulation Mode with a Binding table 1631 6.6.3. The IPv6 Translation-Based Mode 1633 6.6.3.1. Context and Conditions 1635 This mode assumes that IPv6-only terminals are deployed behind port- 1636 restricted devices. Particularly, all DNS resolution requests are 1637 AAAA ones [RFC3363] . Only IPv6 addresses are conveyed in DNS 1638 responses to requesting machines. 1640 A dedicated ALG should be supported by the DNS infrastructure 1641 deployed by the service provider. The main function of this ALG is 1642 to form an IPv6 address based on a Pref6 prefix and a resolved IPv4 1643 address, when no AAAA RR are available in the DNS system (following 1644 the formalism described in Section 4.2). The Pref6 prefix is 1645 configured by the service provider so as to identify that the 1646 resulting IPv6 address is not a native one. We refer to this IPv6 1647 prefix as Pref6_v4. The procedure is illustrated in Figure 18. 1649 +--+ +---+ +---+ 1650 |M1| |CPE| |DNS| 1651 +--+ +---+ +---+ 1652 | | | 1653 |==(1)Request(AAAA)==>|==(2)Request(AAAA)==>| 1654 | | | 1655 | | +------ALG------+ 1656 | | |No available | 1657 | | |AAAA Records, | 1658 | | +------| |------+ 1659 | | | | | | 1660 | | +-------v-------+ 1661 | | | Return IPv4@ | 1662 | | | Build IPv6@ | 1663 | | +-------+-------+ 1664 | | | 1665 |<=(4)Response(IPv6@)=|<=(3)Response(IPv6@)=| 1666 | | | 1668 Figure 18: DNS ALG 1670 In the remaining part of this section, it is assumed that M1 has 1671 retrieved an IPv6 address to contact. 1673 6.6.3.2. Flow Example 1675 Figure 19 shows a message exchange that occurs in the context of 1676 IPv6-IPv4 communications. 1678 +--+ +---+ +-----+ +--+ 1679 |M1| |CPE| |i-PRR| |RM| 1680 +--+ +---+ +-----+ +--+ 1681 | | | | 1682 |==(1)IPv6_Out==>|==(2)IPv6_Out===>|==(3)Pub_IPv4_Out==>| 1683 | | | | 1684 |<==(6)IPv6_In===|<==(5)IPv6_In====|<===(4)Pub_IPv4_In==| 1685 | | | | 1687 Figure 19: Translation Mode 1689 Intra-domain communications are placed using IPv6 transfer 1690 capabilities. When the remote destination is an IPv4 (which is 1691 represented by an IPv6 address), the following exchanges are 1692 observed: 1694 1. M1 issues an IPv6 message destined to RM. 1696 2. Once received by the CPE, this latter checks if the destination 1697 address belongs to the Pref6_v4 prefix. If this is the case, a 1698 NAT66 operation is executed. As a result, a new IPv6 packet is 1699 generated. 1701 3. This message is received by the interconnection PRR. It 1702 retrieves IPv4 information based on IPv6 one and translates the 1703 packet to a new IPv4 one. 1705 4. This message is then routed using IPv4 capabilities of the 1706 connected IPv4-only realm. 1708 5. Once received by RM, an answer may be issued. An IPv4 packet is 1709 then sent. 1711 6. This IPv4 packet is received by i-PRR. It then proceeds to a 1712 stateless NAT46 operation. The newly built IPv6 packet is 1713 forwarded to the next hop. 1715 7. Owing to underlying IGP configuration, the packet is received by 1716 the appropriate CPE which checks its NAT66 table. 1718 8. Because a session has been instantiated, a NAT66 operation is 1719 executed. The resulting IPv6 packet is then received by M1. 1721 6.6.3.3. Provisioning Operations 1722 6.6.3.3.1. IP Connectivity Information 1724 Unlike previous steps, no IPv4 connectivity is provided to customers. 1725 None IPv4 packets are sent, neither by the end-user's device within 1726 the LAN nor by the CPE itself. 1728 IP connectivity is exclusively offered owing to IPv6 transfer 1729 capabilities. Thus, no IPv4 connectivity information is conveyed to 1730 end-user's device. In the meantime, an IPv6 prefix (IPv6Pref) is 1731 assigned to the end-user device (CPE or terminal). This assigned 1732 IPv6 prefix follows the constraints listed in Section 4. 1734 As already mentioned in Section 6.3, the service provider may 1735 allocate to the customer's device a second prefix IPv6 prefix which 1736 is not IPv4-mapped. 1738 6.6.3.3.2. Provisioning Procedure 1740 In addition to what has been mentioned for Step_1 (IPv6 part), a 1741 specific policy should be installed so as to "guide" the behaviour of 1742 the NAT66 function introduced in "Handling Outgoing Traffic" section. 1744 This specific policy needs to be aware of the port range allocated to 1745 the port-restricted device. It is for further study to defined how 1746 the port range can be allocated through IPv6 means (e.g., through a 1747 new DHCPv6 option). 1749 6.6.3.4. Port Restricted Device's Behaviour and Supported Functions 1751 6.6.3.4.1. Port Range Restriction 1753 The port restriction is applied only if the destination IPv6 address 1754 belongs to the Pref6_v4 prefix. Otherwise, no port restriction is 1755 enforced, since it is assumed to be a native IPv6 communication. 1757 A new NAT66 function should be supported by the CPE. This NAT66 is 1758 not required to be supported if a directly connected terminal is 1759 used. But then, its address selection process should follow the 1760 recommendations listed in "Handling Outgoing Traffic" sub-section. 1762 6.6.3.4.2. Handling Outgoing Traffic 1764 The following procedure is applied: 1766 o If the destination IPv6 address belongs to the Pref6_v4 prefix 1767 (this means the destination is not a native IPv6 host and an IPv6- 1768 IPv4 interconnection node will be crossed in the delivery path), 1769 the port-restricted device proceeds to NAT66 operations. 1771 Concretely: 1773 * A port number from the Port Range is selected, this port 1774 replaces the original source port number in the transport part 1775 of the received IPv6 packet; 1777 * A source IPv6 address IPv6PrefA is selected under IPv6Pref (see 1778 Section 4) in such a way that the port value contained in the 1779 port part of this IPv6PrefA address is equal to the selected 1780 port number; 1782 * The received IPv6 (from a machine in the LAN) packet is then 1783 translated to a new IPv6 one with the newly built IPv6PrefA 1784 address as source address and the newly selected source port 1785 number in transport part. 1787 o Otherwise, the packet is forwarded to the next IPv6 hop. 1789 6.6.3.4.3. Handling Incoming Traffic 1791 The following procedure is applied: 1793 o If the source IPv6 address belongs to the Pref6_v4 prefix, the 1794 port-restricted device proceeds to NAT66 operations according to 1795 its active NAT sessions. 1797 o Otherwise, the packet is forwarded to the next IPv6 hop in the 1798 LAN. 1800 6.6.3.5. PRR Behaviour 1802 6.6.3.5.1. Supported Functions 1804 Unlike previous steps, no encapsulation function is required to be 1805 supported by the PRR. Nevertheless, a stateless IPv6-IPv4 (and vice 1806 versa) translation must be supported. 1808 6.6.3.5.2. Localization 1810 No PRRs are required anymore to be maintained in the access segment. 1811 Only PRRs located in the interconnection segment should be deployed. 1812 These nodes should be near ASBRs used to interconnect with IPv4-only 1813 realms. 1815 6.6.3.5.3. Behaviour 1817 The behaviour of the PRR is as follows: 1819 1. Traffic received from an IPv4-only realm: The PRR extracts 1820 destination IPv4 address, source IPv4 address, destination port 1821 number and source port number. A new IPv6 packet is generated 1822 following these rules: 1824 * The destination and source port numbers of the generated 1825 packet are the same as the original IPv4 one. 1827 * The destination IPv6 address follows the formalisms described 1828 in Section 4.2 using a Pref6 configured by the service 1829 provider. 1831 * The source IPv6 address follows the formalism described in 1832 Section 4.2 using a Pref6 provided by the service provider. 1834 2. Traffic destined to an IPv4-only realm: A new IPv4 packet is 1835 generated according to these rules: 1837 * The destination and source port numbers of the generated 1838 packet are the same as the original IPv6 one. 1840 * The destination IPv4 address is extracted from the destination 1841 IPv6 address of the received IPv6 packet. See Section 4.3 for 1842 more information about the used IPv6-to-IPv4 address mapping 1843 function. 1845 * The source IPv4 address is extracted from the source IPv6 1846 address of the received IPv6 packet using the IPv6-to-IPv4 1847 address mapping function defined in Section 4.3. 1849 6.6.3.6. Routing Considerations 1851 IPv4 IGP protocols are not anymore enabled in the core network. Only 1852 IPv6 routing table is maintained by involved routers. 1854 Inter-domain IPv4 connectivity is maintained with IPv4-only realms. 1855 IPv4 network prefixes are mapped to IPv6 prefixes (using a Pref6 1856 prefix provided by the local service provider) which are injected in 1857 the IPv6 IGP deployed protocol. 1859 6.7. Analysis and IPv6 Migration Scenarios 1861 As aforementioned, the deployment of IPv6 is not a problem per se. 1862 The main issue is how to ensure a smooth interconnection between IPv4 1863 and IPv6 realms. interconnection functions and procedures should be 1864 deployed. Currently proposed mechanisms rely on stateful 1865 interconnection nodes (e.g., NAT44 CGN or DS-lite CGN) or requires 1866 that dual-stack nodes (including end hosts and intermediary service 1867 nodes) are deployed everywhere. The first category suffers from a 1868 performance issue and the second one is not realistic approach (since 1869 the adoption of IPv6 may require several years). 1871 This document presents solutions which solve the problem of IPv4 1872 address shortage and which prepare the migration to IPv6. As 1873 described in previous sections, three steps have been identified and 1874 specified. Table 1 gives an overview of the supported IP version per 1875 network segment and for each step. The proposed solution requires 1876 the migration of core segments to IPv6. Dual stacks would be 1877 maintained only at interconnection segments. Table 2 presents the 1878 ratio of IPv6 traffic when the solution is deployed. This table 1879 illustrates the invocation of IPv6 capabilities for the delivery of 1880 IP connectivity service. Owing to the deployment of the proposed 1881 solution, service providers have deterministic means to increase IPv6 1882 traffic. 1884 +--------+--------------+--------------+-----------------+ 1885 | Step | Access | Core | Interconnection | 1886 +--------+--------------+--------------+-----------------+ 1887 | Step_0 | DS Network | IPv4 Network | IPv4 Network | 1888 | Step_1 | DS Network | DS Network | DS Network | 1889 | Step_2 | IPv6 Network | IPv6 Network | DS Network | 1890 +--------+--------------+--------------+-----------------+ 1892 Table 1: Supported IP version per network segment 1894 +------------------------+---------------------+--------------------+ 1895 | Step | %IPv6 traffic | %IPv6 traffic core | 1896 | | Access | | 1897 +------------------------+---------------------+--------------------+ 1898 | Step_0 | at least 50% | variable | 1899 | Step_1 | at least 50% | at least 50% | 1900 | Step_2 (Encapsulation) | 100% | 100% | 1901 | Step_2 (Translation) | 100% | 100% | 1902 +------------------------+---------------------+--------------------+ 1904 Table 2: % of IPv6 1906 Table 3 lists the required function/node to be supported in order to 1907 ensure heterogeneous communication involving peers located in both 1908 IPv4 and IPv6 realms. Core network segment does not require the 1909 deployment of non IPv6 standards elements. Stateless functions 1910 (denoted as PRR in this document) are introduced first at access 1911 segment and then in the interconnection segment. Intra-domain 1912 communications are optimised from an IGP perspective. The presence 1913 of a-PRR allows a natural traffic distribution among deployed nodes. 1914 IPv4 traffic received from adjacent domains is handled by the i-PRR 1915 and then routed to its final destination possibly through the a-PRR 1916 depending of the configuration (binding table with one line per port- 1917 restricted device or stateless mapping between shared IPv4 address 1918 and IPv6 prefixes). 1920 +------------------------+------------+------+-----------------+ 1921 | Step | Access | Core | Interconnection | 1922 +------------------------+------------+------+-----------------+ 1923 | Step_0 | a-PRR | None | None | 1924 | Step_1 | a-PRR | None | i-PRR | 1925 | Step_2 (Encapsulation) | a-PRR/None | None | i-PRR | 1926 | Step_2 (Translation) | None | None | i-PRR | 1927 +------------------------+------------+------+-----------------+ 1929 Table 3: Required nodes per network segment 1931 Table 4 lists the required functions to be enabled in the context of 1932 each step. 1934 +-----------------+-------------------------+-----------------------+ 1935 | Step | port-restricted device | a/i-PRR | 1936 +-----------------+-------------------------+-----------------------+ 1937 | Step_0 | Port Restricted IPv4, | Stateless | 1938 | | IPv4-in-IPv6 | IPv4-in-IPv6 | 1939 | | de-encapsulation | encapsulation | 1940 | Step_1 | Port Restricted IPv4, | stateless | 1941 | | IPv4-in-IPv6 | IPv4-in-IPv6 | 1942 | | de-encapsulation | encapsulation | 1943 | Step_2 | Port Restricted IPv4, | Stateless | 1944 | (Encapsulation) | IPv4-in-IPv6 | IPv4-in-IPv6 | 1945 | | encapsulation, | encapsulation, | 1946 | | IPv4-in-IPv6 | IPv4-in-IPv6 | 1947 | | de-encapsulation | de-encapsulation | 1948 | Step_2 | Port Restricted NAT66 | Stateless NAT46/NAT64 | 1949 | (Translation) | | | 1950 +-----------------+-------------------------+-----------------------+ 1952 Table 4: Required Functions 1954 Various migration paths may be adopted by service providers based on 1955 backward compatibility considerations and also to the service 1956 portfolio. 1958 For service providers which offer already an IPv4-based connectivity 1959 service, several migration paths may be followed, depending on the 1960 service provider's objectives and profile, to adopt IPv6 without 1961 breaking global connectivity (i.e., Reach both IPv4 and IPv6 realms): 1963 1. Deploy IPv6 with no major risks on currently offered services: a 1964 candidate migration path would be (ordered steps): 1966 A. Deploy first the procedure described in 1967 [I-D.boucadair-port-range]. Then, IPv6 may be activated in 1968 the access segment when deploying Step_0. Once IPv6 is 1969 deployed in core network, the service provider should 1970 activate Step_1 mainly by deploying PRR at interconnection 1971 segments. Once the connectivity service is stable, a final 1972 step would be to adopt Step_2 (encapsulation mode) and then 1973 Step_2 (translation mode). 1975 B. Deploy first Step_0, then adopt Step_1. Once the service is 1976 stable, move to Step_2 (encapsulation mode) and latter to 1977 Step_2 (translation mode). 1979 2. Aggressive position with regards to IPv6 deployment: For this 1980 category of service providers, the migration path would be either 1982 A. either deploy Step_1 then Step_2 (encapsulation mode) and 1983 finally Step_2 (translation mode). 1985 B. or deploy Step_2 (encapsulation mode) and then Step_2 1986 (translation mode). 1988 For new service providers which do not have backward compatibility 1989 requirement, the following deployment path may be adopted to ensure a 1990 global IPv4/IPv6 connectivity service. 1992 Deploy Step_2 (encapsulation mode) and then migrate to Step_2 1993 (translation mode). 1995 7. IANA Considerations 1997 TBC. 1999 8. Security Considerations 2001 TBC. 2003 9. Acknowledgements 2005 The authors would like to thank Pierrick MORAND and Eric BURGEY for 2006 their review, support and suggestions. 2008 10. References 2010 10.1. Normative References 2012 [RFC2119] Bradner, S., "Key words for use in RFCs to Indicate 2013 Requirement Levels", BCP 14, RFC 2119, March 1997. 2015 [RFC3315] Droms, R., Bound, J., Volz, B., Lemon, T., Perkins, C., 2016 and M. Carney, "Dynamic Host Configuration Protocol for 2017 IPv6 (DHCPv6)", RFC 3315, July 2003. 2019 10.2. Informative References 2021 [I-D.bajko-pripaddrassign] 2022 Bajko, G., Savolainen, T., Boucadair, M., and P. Levis, 2023 "Port Restricted IP Address Assignment", 2024 draft-bajko-pripaddrassign-01 (work in progress), 2025 March 2009. 2027 [I-D.boucadair-dhcpv6-shared-address-option] 2028 Boucadair, M., Levis, P., Grimault, J., Savolainen, T., 2029 and G. Bajko, "Dynamic Host Configuration Protocol 2030 (DHCPv6) Options for Shared IP Addresses Solutions", 2031 draft-boucadair-dhcpv6-shared-address-option-00 (work in 2032 progress), May 2009. 2034 [I-D.boucadair-port-range] 2035 Boucadair, M., Levis, P., Bajko, G., and T. Savolainen, 2036 "IPv4 Connectivity Access in the Context of IPv4 Address 2037 Exhaustion: Port Range based IP Architecture", 2038 draft-boucadair-port-range-02 (work in progress), 2039 July 2009. 2041 [I-D.boucadair-pppext-portrange-option] 2042 Boucadair, M., Levis, P., Grimault, J., and A. 2043 Villefranque, "Port Range Configuration Options for PPP 2044 IPCP", draft-boucadair-pppext-portrange-option-01 (work in 2045 progress), July 2009. 2047 [I-D.ietf-softwire-dual-stack-lite] 2048 Durand, A., Droms, R., Haberman, B., Woodyatt, J., Lee, 2049 Y., and R. Bush, "Dual-stack lite broadband deployments 2050 post IPv4 exhaustion", 2051 draft-ietf-softwire-dual-stack-lite-01 (work in progress), 2052 July 2009. 2054 [I-D.levis-behave-ipv4-shortage-framework] 2055 Levis, P., Boucadair, M., Grimault, J., and A. 2057 Villefranque, "IPv4 Address Shortage: Needs and Open 2058 Issues", draft-levis-behave-ipv4-shortage-framework-02 2059 (work in progress), June 2009. 2061 [RFC3363] Bush, R., Durand, A., Fink, B., Gudmundsson, O., and T. 2062 Hain, "Representing Internet Protocol version 6 (IPv6) 2063 Addresses in the Domain Name System (DNS)", RFC 3363, 2064 August 2002. 2066 Authors' Addresses 2068 Mohamed Boucadair (editor) 2069 France Telecom 2070 3, Av Francois Chateau 2071 Rennes 35000 2072 France 2074 Email: mohamed.boucadair@orange-ftgroup.com 2076 Pierre Levis 2077 France Telecom 2079 Email: pierre.levis@orange-ftgroup.com 2081 Jean-Luc Grimault 2082 France Telecom 2084 Email: jeanluc.grimault@orange-ftgroup.com 2086 Alain Villefranque 2087 France Telecom 2089 Email: alain.villefranque@orange-ftgroup.com 2091 Mohamed Kassi-Lahlou 2092 France Telecom 2094 Email: mohamed.kassilahlou@orange-ftgroup.com 2095 Gabor Bajko 2096 Nokia 2097 USA 2099 Email: Gabor.Bajko@nokia.com 2100 URI: 2102 Yiu L. Lee 2103 Comcast 2104 U.S.A. 2106 Email: yiu_lee@cable.comcast.com 2107 URI: http://www.comcast.com 2109 Telemaco Melia 2110 Alcatel-Lucent 2111 France 2113 Email: Telemaco.Melia@alcatel-lucent.com 2114 URI: