Debian Codegraph @

rfc6269.html

raw
1 <pre>Internet Engineering Task Force (IETF) M. Ford, Ed.
2 Request for Comments: 6269 Internet Society
3 Category: Informational M. Boucadair
4 ISSN: 2070-1721 France Telecom
5 A. Durand
6 Juniper Networks
7 P. Levis
8 France Telecom
9 P. Roberts
10 Internet Society
11 June 2011
12
13
14 <span class="h1">Issues with IP Address Sharing</span>
15
16 Abstract
17
18 The completion of IPv4 address allocations from IANA and the Regional
19 Internet Registries (RIRs) is causing service providers around the
20 world to question how they will continue providing IPv4 connectivity
21 service to their subscribers when there are no longer sufficient IPv4
22 addresses to allocate them one per subscriber. Several possible
23 solutions to this problem are now emerging based around the idea of
24 shared IPv4 addressing. These solutions give rise to a number of
25 issues, and this memo identifies those common to all such address
26 sharing approaches. Such issues include application failures,
27 additional service monitoring complexity, new security
28 vulnerabilities, and so on. Solution-specific discussions are out of
29 scope.
30
31 Deploying IPv6 is the only perennial way to ease pressure on the
32 public IPv4 address pool without the need for address sharing
33 mechanisms that give rise to the issues identified herein.
34
35 Status of This Memo
36
37 This document is not an Internet Standards Track specification; it is
38 published for informational purposes.
39
40 This document is a product of the Internet Engineering Task Force
41 (IETF). It represents the consensus of the IETF community. It has
42 received public review and has been approved for publication by the
43 Internet Engineering Steering Group (IESG). Not all documents
44 approved by the IESG are a candidate for any level of Internet
45 Standard; see <a href="./rfc5741#section-2">Section&nbsp;2 of RFC 5741</a>.
46
47
48
49
50
51
52 <span class="grey">Ford, et al. Informational [Page 1]</span></pre>
53 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-2" ></span>
54 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
55
56
57 Information about the current status of this document, any errata,
58 and how to provide feedback on it may be obtained at
59 <a href="http://www.rfc-editor.org/info/rfc6269">http://www.rfc-editor.org/info/rfc6269</a>.
60
61 Copyright Notice
62
63 Copyright (c) 2011 IETF Trust and the persons identified as the
64 document authors. All rights reserved.
65
66 This document is subject to <a href="https://www.rfc-editor.org/bcp/bcp78">BCP 78</a> and the IETF Trust's Legal
67 Provisions Relating to IETF Documents
68 (<a href="http://trustee.ietf.org/license-info">http://trustee.ietf.org/license-info</a>) in effect on the date of
69 publication of this document. Please review these documents
70 carefully, as they describe your rights and restrictions with respect
71 to this document. Code Components extracted from this document must
72 include Simplified BSD License text as described in Section 4.e of
73 the Trust Legal Provisions and are provided without warranty as
74 described in the Simplified BSD License.
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108 <span class="grey">Ford, et al. Informational [Page 2]</span></pre>
109 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-3" ></span>
110 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
111
112
113 Table of Contents
114
115 <a href="#section-1">1</a>. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-4">4</a>
116 <a href="#section-2">2</a>. Shared Addressing Solutions . . . . . . . . . . . . . . . . . <a href="#page-4">4</a>
117 3. Analysis of Issues as They Relate to First and Third
118 Parties . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-6">6</a>
119 <a href="#section-4">4</a>. Content Provider Example . . . . . . . . . . . . . . . . . . . <a href="#page-8">8</a>
120 <a href="#section-5">5</a>. Port Allocation . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-8">8</a>
121 <a href="#section-5.1">5.1</a>. Outgoing Ports . . . . . . . . . . . . . . . . . . . . . . <a href="#page-9">9</a>
122 <a href="#section-5.2">5.2</a>. Incoming Ports . . . . . . . . . . . . . . . . . . . . . . <a href="#page-10">10</a>
123 <a href="#section-5.2.1">5.2.1</a>. Port Negotiation . . . . . . . . . . . . . . . . . . . <a href="#page-11">11</a>
124 <a href="#section-5.2.2">5.2.2</a>. Connection to a Well-Known Port Number . . . . . . . . <a href="#page-12">12</a>
125 <a href="#section-5.2.3">5.2.3</a>. Port Discovery Mechanisms . . . . . . . . . . . . . . <a href="#page-12">12</a>
126 <a href="#section-6">6</a>. Impact on Applications . . . . . . . . . . . . . . . . . . . . <a href="#page-12">12</a>
127 <a href="#section-7">7</a>. Geo-location and Geo-proximity . . . . . . . . . . . . . . . . <a href="#page-14">14</a>
128 <a href="#section-8">8</a>. Tracking Service Usage . . . . . . . . . . . . . . . . . . . . <a href="#page-15">15</a>
129 <a href="#section-9">9</a>. ICMP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-15">15</a>
130 <a href="#section-10">10</a>. MTU . . . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-16">16</a>
131 <a href="#section-11">11</a>. Fragmentation . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-16">16</a>
132 <a href="#section-12">12</a>. Traceability . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-17">17</a>
133 <a href="#section-13">13</a>. Security . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-18">18</a>
134 <a href="#section-13.1">13.1</a>. Abuse Logging and Penalty Boxes . . . . . . . . . . . . . <a href="#page-18">18</a>
135 <a href="#section-13.2">13.2</a>. Authentication . . . . . . . . . . . . . . . . . . . . . . <a href="#page-19">19</a>
136 <a href="#section-13.3">13.3</a>. Spam . . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-19">19</a>
137 <a href="#section-13.4">13.4</a>. Port Randomization . . . . . . . . . . . . . . . . . . . . <a href="#page-19">19</a>
138 <a href="#section-13.5">13.5</a>. IPsec . . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
139 <a href="#section-13.6">13.6</a>. Policing Forwarding Behavior . . . . . . . . . . . . . . . <a href="#page-20">20</a>
140 <a href="#section-14">14</a>. Transport Issues . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
141 <a href="#section-14.1">14.1</a>. Parallel Connections . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
142 <a href="#section-14.2">14.2</a>. Serial Connections . . . . . . . . . . . . . . . . . . . . <a href="#page-20">20</a>
143 <a href="#section-14.3">14.3</a>. TCP Control Block Sharing . . . . . . . . . . . . . . . . <a href="#page-21">21</a>
144 <a href="#section-15">15</a>. Reverse DNS . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-21">21</a>
145 <a href="#section-16">16</a>. Load Balancing . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-21">21</a>
146 <a href="#section-17">17</a>. IPv6 Transition Issues . . . . . . . . . . . . . . . . . . . . <a href="#page-21">21</a>
147 <a href="#section-18">18</a>. Introduction of Single Points of Failure . . . . . . . . . . . <a href="#page-22">22</a>
148 <a href="#section-19">19</a>. State Maintenance Reduces Battery Life . . . . . . . . . . . . <a href="#page-22">22</a>
149 <a href="#section-20">20</a>. Support of Multicast . . . . . . . . . . . . . . . . . . . . . <a href="#page-22">22</a>
150 <a href="#section-21">21</a>. Support of Mobile-IP . . . . . . . . . . . . . . . . . . . . . <a href="#page-22">22</a>
151 <a href="#section-22">22</a>. Security Considerations . . . . . . . . . . . . . . . . . . . <a href="#page-23">23</a>
152 <a href="#section-23">23</a>. Contributors . . . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-23">23</a>
153 <a href="#section-24">24</a>. Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . <a href="#page-23">23</a>
154 <a href="#section-25">25</a>. Informative References . . . . . . . . . . . . . . . . . . . . <a href="#page-23">23</a>
155 <a href="#appendix-A">Appendix A</a>. Classes of Address Sharing Solution . . . . . . . . . <a href="#page-27">27</a>
156 <a href="#appendix-B">Appendix B</a>. Address Space Multiplicative Factor . . . . . . . . . <a href="#page-27">27</a>
157
158
159
160
161
162
163
164 <span class="grey">Ford, et al. Informational [Page 3]</span></pre>
165 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-4" ></span>
166 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
167
168
169 <span class="h2"><a class="selflink" id="section-1" href="#section-1">1</a>. Introduction</span>
170
171 Allocations of IPv4 addresses from the Internet Assigned Numbers
172 Authority (IANA) were completed on February 3, 2011 [<a href="#ref-IPv4_Pool">IPv4_Pool</a>].
173 Allocations from Regional Internet Registries (RIRs) are anticipated
174 to be complete around a year later, although the exact date will vary
175 from registry to registry. This is causing service providers around
176 the world to start to question how they will continue providing IPv4
177 connectivity service to their subscribers when there are no longer
178 sufficient IPv4 addresses to allocate them one per subscriber.
179 Several possible solutions to this problem are now emerging based
180 around the idea of shared IPv4 addressing. These solutions give rise
181 to a number of issues, and this memo identifies those common to all
182 such address sharing approaches and collects them in a single
183 document.
184
185 Deploying IPv6 is the only perennial way to ease pressure on the
186 public IPv4 address pool without the need for address sharing
187 mechanisms that give rise to the issues identified herein. In the
188 short term, maintaining growth of IPv4 services in the presence of
189 IPv4 address depletion will require address sharing. Address sharing
190 will cause issues for end-users, service providers, and third parties
191 such as law enforcement agencies and content providers. This memo is
192 intended to highlight and briefly discuss these issues.
193
194 Increased IPv6 deployment should reduce the burden being placed on an
195 address sharing solution, and should reduce the costs of operating
196 that solution. Increasing IPv6 deployment should cause a reduction
197 in the number of concurrent IPv4 sessions per subscriber. If the
198 percentage of end-to-end IPv6 traffic significantly increases, so
199 that the volume of IPv4 traffic begins decreasing, then the number of
200 IPv4 sessions will decrease. The smaller the number of concurrent
201 IPv4 sessions per subscriber, the higher the number of subscribers
202 able to share the same IPv4 public address, and consequently, the
203 lower the number of IPv4 public addresses required. However, this
204 effect will only occur for subscribers who have both an IPv6 access
205 and a shared IPv4 access. This motivates a strategy to
206 systematically bind a shared IPv4 access to an IPv6 access. It is
207 difficult to foresee to what extent growing IPv6 traffic will reduce
208 the number of concurrent IPv4 sessions, but in any event, IPv6
209 deployment and use should reduce the pressure on the available public
210 IPv4 address pool.
211
212 <span class="h2"><a class="selflink" id="section-2" href="#section-2">2</a>. Shared Addressing Solutions</span>
213
214 In many networks today, a subscriber is provided with a single public
215 IPv4 address at their home or small business. For instance, in fixed
216 broadband access, an IPv4 public address is assigned to each CPE
217
218
219
220 <span class="grey">Ford, et al. Informational [Page 4]</span></pre>
221 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-5" ></span>
222 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
223
224
225 (Customer Premises Equipment). CPEs embed a NAT function that is
226 responsible for translating private IPv4 addresses ([<a href="./rfc1918" title="&quot;Address Allocation for Private Internets&quot;">RFC1918</a>]
227 addresses) assigned to hosts within the local network, to the public
228 IPv4 address assigned by the service provider (and vice versa).
229 Therefore, devices located with the LAN share the single public IPv4
230 address and they are all associated with a single subscriber account
231 and a single network operator.
232
233 A number of proposals currently under consideration in the IETF rely
234 upon the mechanism of multiplexing multiple subscribers' connections
235 over a smaller number of shared IPv4 addresses. This is a
236 significant change. These proposals include Carrier Grade NAT (CGN
237 a.k.a. LSN for Large Scale NAT) [<a href="#ref-LSN-REQS" title="&quot;Common requirements for IP address sharing schemes&quot;">LSN-REQS</a>], Dual-Stack Lite
238 [<a href="#ref-DS-Lite" title="&quot;Dual- Stack Lite Broadband Deployments Following IPv4 Exhaustion&quot;">DS-Lite</a>], NAT64 [<a href="./rfc6145" title="&quot;IP/ICMP Translation Algorithm&quot;">RFC6145</a>] [<a href="./rfc6146" title="&quot;Stateful NAT64: Network Address and Protocol Translation from IPv6 Clients to IPv4 Servers&quot;">RFC6146</a>], Address+Port (A+P) proposals
239 [A+P] [<a href="#ref-PORT-RANGE">PORT-RANGE</a>], and Stateless Address Mapping [<a href="#ref-SAM" title="&quot;Scalable Multihoming across IPv6 Local- Address Routing Zones Global-Prefix/Local-Address Stateless Address Mapping (SAM)&quot;">SAM</a>]. <a href="#appendix-A">Appendix A</a>
240 and <a href="#appendix-B">Appendix B</a> provide a classification of these different types of
241 solutions and discuss some of the design considerations to be borne
242 in mind when deploying large-scale address sharing. Whether we're
243 talking about DS-Lite, A+P, NAT64, or CGN isn't especially important
244 -- it's the view from the outside that matters, and given that, most
245 of the issues identified below apply regardless of the specific
246 address sharing scenario in question.
247
248 In these new proposals, a single public IPv4 address would be shared
249 by multiple homes or small businesses (i.e., multiple subscribers),
250 so the operational paradigm described above would no longer apply.
251 In this document, we refer to this new paradigm as large-scale
252 address sharing. All these proposals extend the address space by
253 adding port information; they differ in the way they manage the port
254 value.
255
256 Security issues associated with NAT have long been documented (see
257 [<a href="./rfc2663" title="&quot;IP Network Address Translator (NAT) Terminology and Considerations&quot;">RFC2663</a>] and [<a href="./rfc2993" title="&quot;Architectural Implications of NAT&quot;">RFC2993</a>]). However, sharing IPv4 addresses across
258 multiple subscribers by any means, either moving the NAT
259 functionality from the home gateway to the core of the service
260 provider network or restricting the port choice in the subscriber's
261 NAT, creates additional issues for subscribers, content providers,
262 and network operators. Many of these issues are created today by
263 public Wi-Fi hotspot deployments. As such large-scale address
264 sharing solutions become more widespread in the face of IPv4 address
265 depletion, these issues will become both more severe and more
266 commonly felt. NAT issues in the past typically only applied to a
267 single legal entity; as large-scale address sharing becomes more
268 prevalent, these issues will increasingly span across multiple legal
269 entities simultaneously.
270
271
272
273
274
275
276 <span class="grey">Ford, et al. Informational [Page 5]</span></pre>
277 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-6" ></span>
278 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
279
280
281 All large-scale address sharing proposals share technical and
282 operational issues, and these are addressed in the sections that
283 follow. These issues are common to any service-provider NAT,
284 enterprise NAT, and also non-NAT solutions that share individual IPv4
285 addresses across multiple subscribers. This document is intended to
286 bring all of these issues together in one place.
287
288 <span class="h2"><a class="selflink" id="section-3" href="#section-3">3</a>. Analysis of Issues as They Relate to First and Third Parties</span>
289
290 In this section, we present an analysis of whether the issues
291 identified in the remainder of this document are applicable to the
292 organization deploying the shared addressing mechanism (and by
293 extension their subscribers) and/or whether these issues impact third
294 parties (e.g., content providers, law enforcement agencies, etc.).
295 In this analysis, issues that affect end-users are deemed to affect
296 first parties, as end-users can be expected to complain to their
297 service provider when problems arise. Where issues can expect to be
298 foreseen and addressed by the party deploying the shared addressing
299 solution, they are not attributed.
300
301 In Figure 1, we have also tried to indicate (with 'xx') where issues
302 are newly created in addition to what could be expected from the
303 introduction of a traditional NAT device. Issues marked with a
304 single 'x' are already present today in the case of typical CPE NAT;
305 however, they can be expected to be more severe and widespread in the
306 case of large-scale address sharing.
307
308 +------------------------------------------------+--------+---------+
309 | Issue | 1st | 3rd |
310 | | party | parties |
311 +------------------------------------------------+--------+---------+
312 | Restricted allocations of outgoing | x | |
313 | ports will impact performance for end-users | | |
314 | | | |
315 | Incoming port negotiation mechanisms may fail | xx | |
316 | | | |
317 | Incoming connections to Assigned Ports will | x | |
318 | not work | | |
319 | | | |
320 | Port discovery mechanisms will not work | x | |
321 | | | |
322 | Some applications will fail to operate | x | x |
323 | | | |
324 | Assumptions about parallel/serial connections | x | x |
325 | may fail | | |
326 | | | |
327 +------------------------------------------------+--------+---------+
328
329
330
331
332 <span class="grey">Ford, et al. Informational [Page 6]</span></pre>
333 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-7" ></span>
334 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
335
336
337 +------------------------------------------------+--------+---------+
338 | Issue | 1st | 3rd |
339 | | party | parties |
340 +------------------------------------------------+--------+---------+
341 | TCP control block sharing will be affected | x | x |
342 | | | |
343 | Reverse DNS will be affected | x | x |
344 | | | |
345 | Inbound ICMP will fail in many cases | x | x |
346 | | | |
347 | Amplification of security issues will occur | xx | xx |
348 | | | |
349 | Fragmentation will require special handling | x | |
350 | | | |
351 | Single points of failure and increased | x | |
352 | network instability may occur | | |
353 | | | |
354 | Port randomization will be affected | x | |
355 | | | |
356 | Service usage monitoring and abuse logging | xx | xx |
357 | will be impacted for all elements in the chain | | |
358 | between service provider and content provider | | |
359 | | | |
360 | Penalty boxes will no longer work | xx | xx |
361 | | | |
362 | Spam blacklisting will be affected | xx | xx |
363 | | | |
364 | Geo-location services will be impacted | xx | xx |
365 | | | |
366 | Geo-proximity mechanisms will be impacted | xx | xx |
367 | | | |
368 | Load balancing algorithms may be impacted | | xx |
369 | | | |
370 | Authentication mechanisms may be impacted | | x |
371 | | | |
372 | Traceability of network usage and abusage will | | xx |
373 | be affected | | |
374 | | | |
375 | IPv6 transition mechanisms will be affected | xx | |
376 | | | |
377 | Frequent keep-alives will reduce battery life | x | |
378 | | | |
379 +------------------------------------------------+--------+---------+
380
381 Figure 1: Shared addressing issues for first and third parties
382
383
384
385
386
387
388 <span class="grey">Ford, et al. Informational [Page 7]</span></pre>
389 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-8" ></span>
390 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
391
392
393 As can be seen from Figure 1, deployment of large-scale address
394 sharing will create almost as many problems for third parties as it
395 does for the service provider deploying the address sharing
396 mechanism. Several of these issues are specific to the introduction
397 of large-scale address sharing as well. All of these issues are
398 discussed in further detail below.
399
400 <span class="h2"><a class="selflink" id="section-4" href="#section-4">4</a>. Content Provider Example</span>
401
402 Taking a content provider as an example of a third party, and
403 focusing on the issues that are created specifically by the presence
404 of large-scale address sharing, we identify the following issues as
405 being of particular concern:
406
407 o Degraded geo-location for targeted advertising and licensed
408 content restrictions (see <a href="#section-7">Section 7</a>).
409
410 o Additional latency due to indirect routing and degraded geo-
411 proximity (see <a href="#section-7">Section 7</a>).
412
413 o Exposure to new amplification attacks (see <a href="#section-13">Section 13</a>).
414
415 o Service usage monitoring is made more complicated (see <a href="#section-8">Section 8</a>).
416
417 o Incoming port negotiation mechanisms may fail (see <a href="#section-5.2.1">Section 5.2.1</a>).
418
419 o IP blocking for abuse/spam will cause collateral damage (see
420 <a href="#section-13">Section 13</a>).
421
422 o Load balancing algorithms may be impacted (see <a href="#section-16">Section 16</a>).
423
424 o Traceability of network usage and abuse will be impacted (see
425 <a href="#section-12">Section 12</a>).
426
427 <span class="h2"><a class="selflink" id="section-5" href="#section-5">5</a>. Port Allocation</span>
428
429 When we talk about port numbers, we need to make a distinction
430 between outgoing connections and incoming connections. For outgoing
431 connections, the actual source port number used is usually
432 irrelevant. (While this is true today, in a port-range solution, it
433 is necessary for the source port to be within the allocated range.)
434 But for incoming connections, the specific port numbers allocated to
435 subscribers matter because they are part of external referrals (used
436 by third parties to contact services run by the subscribers).
437
438 The total number of subscribers able to share a single IPv4 address
439 will depend upon assumptions about the average number of ports
440 required per active subscriber, and the average number of
441
442
443
444 <span class="grey">Ford, et al. Informational [Page 8]</span></pre>
445 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-9" ></span>
446 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
447
448
449 simultaneously active subscribers. It is important to realize that
450 the TCP design makes it undesirable for clients to re-use ports while
451 they remain in the TIME-WAIT state (typically 4 minutes after the
452 connection has concluded). TIME-WAIT state removes the hazard of old
453 duplicates for "fast" or "long" connections, in which clock-driven
454 Initial Sequence Number selection is unable to prevent overlap of the
455 old and new sequence spaces. The TIME-WAIT delay allows all old
456 duplicate segments time enough to die in the Internet before the
457 connection is reopened [<a href="./rfc1337" title="&quot;TIME-WAIT Assassination Hazards in TCP&quot;">RFC1337</a>]. Therefore, ports in this state
458 must be included in calculations concerning port usage per
459 subscriber.
460
461 Most of the time the source port selected by a client application
462 will be translated (unless there is direct knowledge of a port-range
463 restriction in the client's stack), either by a NAT in the
464 subscriber's device, or by a CPE NAT, or by a CPE NAT and a CGN.
465
466 [<a id="ref-RFC1700">RFC1700</a>] (which was replaced by an online database, as described by
467 [<a href="./rfc3232" title="&quot;Assigned Numbers: RFC 1700 is Replaced by an On-line Database&quot;">RFC3232</a>]) defines the Assigned Ports (both System and User). IANA
468 has further classified the whole port space into three categories, as
469 defined in [<a href="#ref-IANA_Ports">IANA_Ports</a>]:
470
471 o The Well-Known Ports are those from 0 through 1023.
472
473 o The Registered Ports are those from 1024 through 49151.
474
475 o The Dynamic and/or Private Ports are those from 49152 through
476 65535.
477
478 [<a id="ref-RFC4787">RFC4787</a>] notes that current NATs have different policies with regard
479 to this classification; some NATs restrict their translations to the
480 use of dynamic ports, some also include registered ports, some
481 preserve the port if it is in the well-known range. [<a href="./rfc4787" title="&quot;Network Address Translation (NAT) Behavioral Requirements for Unicast UDP&quot;">RFC4787</a>] makes
482 it clear that the use of port space (1024-65535) is safe: "mapping a
483 source port to a source port that is already registered is unlikely
484 to have any bad effects". Therefore, for all address sharing
485 solutions, there is no reason to only consider a subset of the port
486 space (1024-65535) for outgoing source ports.
487
488 <span class="h3"><a class="selflink" id="section-5.1" href="#section-5.1">5.1</a>. Outgoing Ports</span>
489
490 According to measurements, the average number of outgoing ports
491 consumed per active subscriber is much, much smaller than the maximum
492 number of ports a subscriber can use at any given time. However, the
493 distribution is heavy-tailed, so there are typically a small number
494 of subscribers who use a very high number of ports [<a href="#ref-CGN_Viability">CGN_Viability</a>].
495 This means that an algorithm that dynamically allocates outgoing port
496 numbers from a central pool will typically allow more subscribers to
497
498
499
500 <span class="grey">Ford, et al. Informational [Page 9]</span></pre>
501 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-10" ></span>
502 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
503
504
505 share a single IPv4 address than algorithms that statically divide
506 the resource by pre-allocating a fixed number of ports to each
507 subscriber. Similarly, such an algorithm should be more able to
508 accommodate subscribers wishing to use a relatively high number of
509 ports.
510
511 It is important to note here that the desire to dynamically allocate
512 outgoing port numbers will make a service provider's job of
513 maintaining records of subscriber port number allocations
514 considerably more onerous (see <a href="#section-12">Section 12</a>). The number of records
515 per subscriber will increase from 1 in a scheme where ports are
516 statically allocated, to a much larger number equivalent to the total
517 number of outgoing ports consumed by that subscriber during the time
518 period for which detailed logs must be kept.
519
520 A potential problem with dynamic allocation occurs when one of the
521 subscriber devices behind such a port-shared IPv4 address becomes
522 infected with a worm, which then quickly sets about opening many
523 outbound connections in order to propagate itself. Such an infection
524 could rapidly exhaust the shared resource of the single IPv4 address
525 for all connected subscribers. It is therefore necessary to impose
526 limits on the total number of ports available to an individual
527 subscriber to ensure that the shared resource (the IPv4 address)
528 remains available in some capacity to all the subscribers using it.
529 However, static schemes for ports assignment may introduce security
530 issues [<a href="./rfc6056" title="&quot;Recommendations for Transport- Protocol Port Randomization&quot;">RFC6056</a>] when small contiguous port ranges are statically
531 assigned to subscribers. Another way to mitigate this issue is to
532 kill off (randomly) established connections when the port space runs
533 out. A user with many connections will be proportionally more likely
534 to get impacted.
535
536 Session failure due to NAT state overflow or timeout (when the NAT
537 discards session state because it's run out of resource) can be
538 experienced when the configured quota per user is reached or if the
539 NAT is out of resources.
540
541 <span class="h3"><a class="selflink" id="section-5.2" href="#section-5.2">5.2</a>. Incoming Ports</span>
542
543 It is desirable to ensure that incoming ports remain stable over
544 time. This is challenging as the network doesn't know anything in
545 particular about the applications that it is supporting, and
546 therefore has no real notion of how long an application/service
547 session is still ongoing and therefore requiring port stability.
548
549 Early measurements [<a href="#ref-CGN_Viability">CGN_Viability</a>] also seem to indicate that, on
550 average, only very few ports are used by subscribers for incoming
551 connections. However, a majority of subscribers accept at least one
552 inbound connection.
553
554
555
556 <span class="grey">Ford, et al. Informational [Page 10]</span></pre>
557 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-11" ></span>
558 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
559
560
561 This means that it is not necessary to pre-allocate a large number of
562 incoming ports to each subscriber. It is possible to either pre-
563 allocate a small number of ports for incoming connections or do port
564 allocation on demand when the application wishing to receive a
565 connection is initiated. The bulk of incoming ports can be reserved
566 as a centralized resource shared by all subscribers using a given
567 public IPv4 address.
568
569 <span class="h4"><a class="selflink" id="section-5.2.1" href="#section-5.2.1">5.2.1</a>. Port Negotiation</span>
570
571 In current deployments, one important and widely used feature of many
572 CPE devices is the ability to open incoming ports (port forwarding)
573 either manually, or with a protocol such as the Universal Plug and
574 Play Internet Gateway Device (UPnP IGD) [<a href="#ref-UPnP-IGD" title="&quot;Universal Plug and Play (UPnP) Internet Gateway Device (IGD) V 2.0&quot;">UPnP-IGD</a>]. If a CGN is
575 present, the port must also be opened in the CGN. CGN makes
576 subscribers dependent on their service provider for this
577 functionality.
578
579 CPE and CGN will need to cooperate in order for port forwarding
580 functionality to work. UPnP, or NAT-PMP proxy could be a solution if
581 there is a direct link (or tunnel) between the CPE and the CGN. An
582 alternative solution is a web interface to configure the incoming
583 port mapping on the CGN. Protocol development is underway in the
584 IETF to provide a generalized, automated solution via the Port
585 Control Protocol [<a href="#ref-PCP" title="&quot;Port Control Protocol (PCP)&quot;">PCP</a>].
586
587 Note that such an interface effectively makes public what was
588 previously a private management interface and this raises security
589 concerns that must be addressed.
590
591 For port-range solutions, port forwarding capabilities may still be
592 present at the CPE, with the limitation that the open incoming port
593 must be within the allocated port range (for instance, it is not
594 possible to open port 5002 for incoming connections if port 5002 is
595 not within the allocated port range).
596
597 <span class="h5"><a class="selflink" id="section-5.2.1.1" href="#section-5.2.1.1">5.2.1.1</a>. Universal Plug and Play (UPnP)</span>
598
599 Using the UPnP semantic, an application asks "I want to use port
600 number X, is that OK?", and the answer is yes or no. If the answer
601 is no, the application will typically try the next port in sequence,
602 until it either finds one that works or gives up after a limited
603 number of attempts. UPnP IGD 1.0 has no way to redirect the
604 application to use another port number. UPnP IGD 2.0 improves this
605 situation and allows for allocation of any available port.
606
607
608
609
610
611
612 <span class="grey">Ford, et al. Informational [Page 11]</span></pre>
613 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-12" ></span>
614 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
615
616
617 <span class="h5"><a class="selflink" id="section-5.2.1.2" href="#section-5.2.1.2">5.2.1.2</a>. NAT Port Mapping Protocol (NAT-PMP)</span>
618
619 NAT-PMP enables the NAT to redirect the requesting application to a
620 port deemed to be available for use by the NAT state mapping table.
621
622 <span class="h4"><a class="selflink" id="section-5.2.2" href="#section-5.2.2">5.2.2</a>. Connection to a Well-Known Port Number</span>
623
624 Once an IPv4-address sharing mechanism is in place, inbound
625 connections to well-known port numbers will not work in the general
626 case. Any application that is not port-agile cannot be expected to
627 work. Some workaround (e.g., redirects to a port-specific URI) could
628 be deployed given sufficient incentives. There exist several
629 proposals for 'application service location' protocols that would
630 provide a means of addressing this problem, but historically these
631 proposals have not gained much deployment traction.
632
633 For example, the use of DNS SRV records [<a href="./rfc2782" title="&quot;A DNS RR for specifying the location of services (DNS SRV)&quot;">RFC2782</a>] provides a
634 potential solution for subscribers wishing to host services in the
635 presence of a shared-addressing scheme. SRV records make it possible
636 to specify a port value related to a service, thereby making services
637 accessible on ports other than the well-known ports. It is worth
638 noting that this mechanism is not applicable to HTTP and many other
639 protocols.
640
641 <span class="h4"><a class="selflink" id="section-5.2.3" href="#section-5.2.3">5.2.3</a>. Port Discovery Mechanisms</span>
642
643 Port discovery using a UDP port to discover a service available on a
644 corresponding TCP port, either through broadcast, multicast, or
645 unicast, is a commonly deployed mechanism. Unsolicited inbound UDP
646 will be dropped by address sharing mechanisms as they have no live
647 mapping to enable them to forward the packet to the appropriate host.
648 Address sharing thereby breaks this service discovery technique.
649
650 <span class="h2"><a class="selflink" id="section-6" href="#section-6">6</a>. Impact on Applications</span>
651
652 Address sharing solutions will have an impact on the following types
653 of applications:
654
655 o Applications that establish inbound communications - These
656 applications will have to ensure that ports selected for inbound
657 communications are either within the allocated range (for port-
658 range solutions) or are forwarded appropriately by the CGN (for
659 CGN-based solutions). See <a href="#section-5.2">Section 5.2</a> for more discussion.
660
661 o Applications that carry address and/or port information in their
662 payload - Where translation of port and/or address information is
663 performed at the IP and transport layers by the address sharing
664 solution, an ALG will also be required to ensure application-layer
665
666
667
668 <span class="grey">Ford, et al. Informational [Page 12]</span></pre>
669 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-13" ></span>
670 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
671
672
673 data is appropriately modified. Note that ALGs are required in
674 some cases, and in many other cases end-to-end protocol mechanisms
675 have developed to work around a lack of ALGs in address
676 translators, to the point of it being preferable to avoid any
677 support in the NAT.
678
679 o Applications that use fixed ports - See <a href="#section-5.2.2">Section 5.2.2</a> for more
680 discussion.
681
682 o Applications that do not use any port (e.g., ICMP echo) - Such
683 applications will require special handling -- see <a href="#section-9">Section 9</a> for
684 more discussion.
685
686 o Applications that assume the uniqueness of source addresses (e.g.,
687 IP address as identifier) - Such applications will fail to operate
688 correctly in the presence of multiple, discrete, simultaneous
689 connections from the same source IP address.
690
691 o Applications that explicitly prohibit concurrent connections from
692 the same address - Such applications will fail when multiple
693 subscribers sharing an IP address attempt to use them
694 simultaneously.
695
696 o Applications that do not use TCP or UDP for transport - All IP-
697 address sharing mechanisms proposed to date are limited to TCP,
698 UDP, and ICMP, thereby preventing end-users from fully utilizing
699 the Internet (e.g., SCTP, DCCP, RSVP, protocol 41 (IPv6-over-
700 IPv4), protocol 50 (IPsec ESP)).
701
702 Applications already frequently implement mechanisms in order to
703 circumvent the presence of NATs (typically CPE NATs):
704
705 o Application Layer Gateways (ALGs): Many CPE devices today embed
706 ALGs that allow applications to behave correctly despite the
707 presence of NAT on the CPE. When the NAT belongs to the
708 subscriber, the subscriber has flexibility to tailor the device to
709 his or her needs. For CGNs, subscribers will be dependent on the
710 set of ALGs that their service provider makes available. For
711 port-range solutions, ALGs will require modification to deal with
712 the port-range restriction, but will otherwise have the same
713 capabilities as today. Note that ALGs are required in some cases,
714 and in many other cases end-to-end protocol mechanisms have
715 developed to work around a lack of ALGs, to the point of it being
716 preferable to avoid any support in the NAT.
717
718 o NAT Traversal Techniques: There are several commonly deployed
719 mechanisms that support operating servers behind a NAT by
720 forwarding specific TCP or UDP ports to specific internal hosts
721
722
723
724 <span class="grey">Ford, et al. Informational [Page 13]</span></pre>
725 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-14" ></span>
726 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
727
728
729 ([<a href="#ref-UPnP-IGD" title="&quot;Universal Plug and Play (UPnP) Internet Gateway Device (IGD) V 2.0&quot;">UPnP-IGD</a>], [<a href="#ref-NAT-PMP" title="&quot;NAT Port Mapping Protocol (NAT-PMP)&quot;">NAT-PMP</a>], and manual configuration). All of these
730 mechanisms assume the NAT's WAN address is a publicly routable IP
731 address, and fail to work normally when that assumption is wrong.
732 There have been attempts to avoid that problem by automatically
733 disabling the NAT function and bridging traffic instead upon
734 assignment of a private IP address to the WAN interface (as is
735 required for [<a href="#ref-Windows-Logo">Windows-Logo</a>] certification). Bridging (rather than
736 NATting) has other side effects (DHCP requests are served by an
737 upstream DHCP server that can increase complexity of in-home
738 networking).
739
740 <span class="h2"><a class="selflink" id="section-7" href="#section-7">7</a>. Geo-location and Geo-proximity</span>
741
742 IP addresses are frequently used to indicate, with some level of
743 granularity and some level of confidence, where a host is physically
744 located. Using IP addresses in this fashion is a heuristic at best,
745 and is already challenged today by other deployed capabilities, e.g.,
746 tunnels. Geo-location services are used by content providers to
747 allow them to conform with regional content licensing restrictions,
748 to target advertising at specific geographic areas, or to provide
749 customized content. Geo-location services are also necessary for
750 emergency services provision. In some deployment contexts (e.g.,
751 centralized CGN), shared addressing will reduce the level of
752 confidence and level of location granularity that IP-based geo-
753 location services can provide. Viewed from the content provider, a
754 subscriber behind a CGN geo-locates to wherever the prefix of the CGN
755 appears to be; very often that will be in a different city than the
756 subscriber.
757
758 IP addresses are also used as input to geo-location services that
759 resolve an IP address to a physical location using information from
760 the network infrastructure. Current systems rely on resources such
761 as RADIUS databases and DHCP lease tables. The use of address
762 sharing will prevent these systems from resolving the location of a
763 host based on IP address alone. It will be necessary for users of
764 such systems to provide more information (e.g., TCP or UDP port
765 numbers), and for the systems to use this information to query
766 additional network resources (e.g., Network Address Translation -
767 Protocol Translation (NAT-PT) binding tables). Since these new data
768 elements tend to be more ephemeral than those currently used for geo-
769 location, their use by geo-location systems may require them to be
770 cached for some period of time.
771
772 Other forms of geo-location will still work as usual.
773
774 A slightly different use of an IP address is to calculate the
775 proximity of a connecting host to a particular service delivery
776 point. This use of IP address information impacts the efficient
777
778
779
780 <span class="grey">Ford, et al. Informational [Page 14]</span></pre>
781 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-15" ></span>
782 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
783
784
785 delivery of content to an end-user. If a CGN is introduced in
786 communications and it is far from an end-user connected to it,
787 application performance may be degraded insofar as IP-based geo-
788 proximity is a factor.
789
790 <span class="h2"><a class="selflink" id="section-8" href="#section-8">8</a>. Tracking Service Usage</span>
791
792 As large-scale address sharing becomes more commonplace, monitoring
793 the number of unique users of a service will become more complex than
794 simply counting the number of connections from unique IP addresses.
795 While this is a somewhat inexact methodology today due to the
796 widespread use of CPE NAT, it will become a much less useful approach
797 in the presence of widespread large-scale address sharing solutions.
798 In general, all elements that monitor usage or abusage in the chain
799 between a service provider that has deployed address sharing and a
800 content provider will need to be upgraded to take account of the port
801 value in addition to IP addresses.
802
803 <span class="h2"><a class="selflink" id="section-9" href="#section-9">9</a>. ICMP</span>
804
805 ICMP does not include a port field and is consequently problematic
806 for address sharing mechanisms. Some ICMP message types include a
807 fragment of the datagram that triggered the signal to be sent, which
808 is assumed to include port numbers. For some ICMP message types, the
809 Identifier field has to be used as a de-multiplexing token. Sourcing
810 ICMP echo messages from hosts behind an address sharing solution does
811 not pose problems, although responses to outgoing ICMP echo messages
812 will require special handling, such as making use of the ICMP
813 Identifier value to route the response appropriately.
814
815 For inbound ICMP there are two cases. The first case is that of ICMP
816 sourced from outside the network of the address sharing solution
817 provider. Where ICMP messages include a fragment of an outgoing
818 packet including port numbers, it may be possible to forward the
819 packet appropriately. In addition to these network signaling
820 messages, several applications (e.g., peer-to-peer applications) make
821 use of ICMP echo messages that include no hints that could be used to
822 route the packet correctly. Measurements derived by such
823 applications in the presence of an address sharing solution will be
824 erroneous or fail altogether. The second case is that of ICMP
825 sourced from within the network of the address sharing solution
826 provider (e.g., for network management, signaling, and diagnostic
827 purposes). In this case, ICMP can be routed normally for CGN-based
828 solutions owing to the presence of locally unique private IP
829 addresses for each CPE device. For port-range solutions, ICMP echo
830 messages will not be routable without special handling, e.g., placing
831 a port number in the ICMP Identifier field, and having port-range
832 routers make routing decisions based upon that field.
833
834
835
836 <span class="grey">Ford, et al. Informational [Page 15]</span></pre>
837 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-16" ></span>
838 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
839
840
841 Considerations related to ICMP message handling in NAT-based
842 environments are specified in [<a href="./rfc5508" title="&quot;NAT Behavioral Requirements for ICMP&quot;">RFC5508</a>].
843
844 <span class="h2"><a class="selflink" id="section-10" href="#section-10">10</a>. MTU</span>
845
846 In applications where the end hosts are attempting to use path MTU
847 Discovery [<a href="./rfc1191" title="&quot;Path MTU discovery&quot;">RFC1191</a>] to optimize transmitted packet size with
848 underlying network MTU, shared addressing has a number of items that
849 must be considered. As covered in <a href="#section-9">Section 9</a>, ICMP "Packet Too Big"
850 messages must be properly translated through the address sharing
851 solution in both directions. However, even when this is done
852 correctly, MTU can be a concern. Many end hosts cache information
853 that was received via Path MTU Discovery (PMTUD) for a certain period
854 of time. If the MTU behind the address sharing solution is
855 inconsistent, the public end host may have the incorrect MTU value
856 cached. This may cause it to send packets that are too large,
857 causing them to be dropped if the DF (Don't Fragment) bit is set, or
858 causing them to be fragmented by the network, increasing load and
859 overhead. Because the host eventually will reduce MTU to the lowest
860 common value for all hosts behind a given public address, it may also
861 send packets that are below optimal size for the specific connection,
862 increasing overhead and reducing throughput.
863
864 This issue also generates a potential attack vector -- a malevolent
865 user could send an ICMP "Packet Too Big" (Type 3, Code 4) message
866 indicating a Next-Hop MTU of anything down to 68 octets. This value
867 will be cached by the off-net server for all subscribers sharing the
868 address of the malevolent user. This could lead to a denial of
869 service (DoS) against both the remote server and the large-scale NAT
870 device itself (as they will both have to handle many more packets per
871 second).
872
873 <span class="h2"><a class="selflink" id="section-11" href="#section-11">11</a>. Fragmentation</span>
874
875 When a packet is fragmented, transport-layer port information (either
876 UDP or TCP) is only present in the first fragment. Subsequent
877 fragments will not carry the port information and so will require
878 special handling. In addition, the IP Identifier may no longer be
879 unique as required by the receiver to aid in assembling the fragments
880 of a datagram.
881
882
883
884
885
886
887
888
889
890
891
892 <span class="grey">Ford, et al. Informational [Page 16]</span></pre>
893 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-17" ></span>
894 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
895
896
897 <span class="h2"><a class="selflink" id="section-12" href="#section-12">12</a>. Traceability</span>
898
899 In many jurisdictions, service providers are legally obliged to
900 provide the identity of a subscriber upon request to the appropriate
901 authorities. Such legal requests have traditionally included the
902 source IPv4 address and date (and usually the time), which is
903 sufficient information when subscribers are assigned IPv4 addresses
904 for a long duration.
905
906 However, where one public IPv4 address is shared between several
907 subscribers, the IPv4 address no longer uniquely identifies a
908 subscriber. There are two solutions to this problem:
909
910 o The first solution is for servers to additionally log the source
911 port of incoming connections and for the legal request to include
912 the source port. The legal request should include the
913 information: [Source IP address, Source Port, Timestamp] (and
914 possibly other information). Accurate time-keeping (e.g., use of
915 NTP or Simple NTP) is vital because port assignments are dynamic.
916 A densely populated CGN could mean even very small amounts of
917 clock skew between a third party's server and the CGN operator
918 will result in ambiguity about which customer was using a specific
919 port at a given time.
920
921 o The second solution considers it unrealistic to expect all servers
922 to log the source port number of incoming connections. To deal
923 with this, service providers using IPv4 address sharing may need
924 to log IP destination addresses.
925
926 Destination logging is imperfect if multiple subscribers are
927 accessing the same (popular) server at nearly the same time; it can
928 be impossible to disambiguate which subscriber accessed the server,
929 especially with protocols that involve several connections (e.g.,
930 HTTP). Thus, logging the destination address on the NAT is inferior
931 to logging the source port at the server.
932
933 If neither solution is used (that is, the server is not logging
934 source port numbers and the NAT is not logging destination IP
935 addresses), the service provider cannot trace a particular activity
936 to a specific subscriber. In this circumstance, the service provider
937 would need to disclose the identity of all subscribers who had active
938 sessions on the NAT during the time period in question. This may be
939 a large number of subscribers.
940
941 Address sharing solutions must record and store all mappings
942 (typically during 6-12 months, depending on the local jurisdiction)
943 that they create. If we consider one mapping per session, a service
944 provider should record and retain traces of all sessions created by
945
946
947
948 <span class="grey">Ford, et al. Informational [Page 17]</span></pre>
949 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-18" ></span>
950 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
951
952
953 all subscribers during one year (if the legal storage duration is one
954 year). This may be challenging due to the volume of data requiring
955 storage, the volume of data to repeatedly transfer to the storage
956 location, and the volume of data to search in response to a query.
957
958 Address sharing solutions may mitigate these issues to some extent by
959 pre-allocating groups of ports. Then only the allocation of the
960 group needs to be recorded, and not the creation of every session
961 binding within that group. There are trade-offs to be made between
962 the sizes of these port groups, the ratio of public addresses to
963 subscribers, whether or not these groups timeout, and the impact on
964 logging requirements and port randomization security [<a href="./rfc6056" title="&quot;Recommendations for Transport- Protocol Port Randomization&quot;">RFC6056</a>].
965
966 <span class="h2"><a class="selflink" id="section-13" href="#section-13">13</a>. Security</span>
967
968 Before noting some specific security-related issues caused by large-
969 scale address sharing, it is perhaps worth noting that, in general,
970 address sharing creates a vector for attack amplification in numerous
971 ways. See <a href="#section-10">Section 10</a> for one example.
972
973 <span class="h3"><a class="selflink" id="section-13.1" href="#section-13.1">13.1</a>. Abuse Logging and Penalty Boxes</span>
974
975 When an abuse is reported today, it is usually done in the form: IPv4
976 address X has done something bad at time T0. This is not enough
977 information to uniquely identify the subscriber responsible for the
978 abuse when that IPv4 address is shared by more than one subscriber.
979 Law enforcement authorities may be particularly impacted because of
980 this. This particular issue can be fixed by logging port numbers,
981 although this will increase logging data storage requirements.
982
983 A number of services on the network today log the IPv4 source
984 addresses used in connection attempts to protect themselves from
985 certain attacks. For example, if a server sees too many requests
986 from the same IPv4 address in a short period of time, it may decide
987 to put that address in a penalty box for a certain time during which
988 requests are denied, or it may require completion of a CAPTCHA
989 (Completely Automated Public Turing test to tell Computers and Humans
990 Apart) for future requests. If an IPv4 address is shared by multiple
991 subscribers, this would have unintended consequences in a couple of
992 ways. First it may become the natural behavior to see many login
993 attempts from the same address because it is now shared across a
994 potentially large number of subscribers. Second and more likely is
995 that one user who fails a number of login attempts may block out
996 other users who have not made any previous attempts but who will now
997 fail on their first attempt. In the presence of widespread large-
998 scale address sharing, penalty box solutions to service abuse simply
999 will not work.
1000
1001
1002
1003
1004 <span class="grey">Ford, et al. Informational [Page 18]</span></pre>
1005 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-19" ></span>
1006 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1007
1008
1009 In addition, there are web tie-ins into different blacklists that web
1010 administrators subscribe to in order to redirect users with infected
1011 machines (e.g., detect the presence of a worm) to a URL that says
1012 "Hey, your machine is infected!". With address sharing, someone
1013 else's worm can interfere with the ability to access the service for
1014 other subscribers sharing the same IP address.
1015
1016 <span class="h3"><a class="selflink" id="section-13.2" href="#section-13.2">13.2</a>. Authentication</span>
1017
1018 Simple address-based identification mechanisms that are used to
1019 populate access control lists will fail when an IP address is no
1020 longer sufficient to identify a particular subscriber. Including
1021 port numbers in access control list definitions may be possible at
1022 the cost of extra complexity, and may also require the service
1023 provider to make static port assignments, which conflicts with the
1024 requirement for dynamic assignments discussed in <a href="#section-5.1">Section 5.1</a>.
1025
1026 Address or DNS-name-based signatures (e.g., some X.509 signatures)
1027 may also be affected by address sharing as the address itself is now
1028 a shared token, and the name to address mapping may not be current.
1029
1030 <span class="h3"><a class="selflink" id="section-13.3" href="#section-13.3">13.3</a>. Spam</span>
1031
1032 Another case of identifying abusers has to do with spam blacklisting.
1033 When a spammer is behind a CGN or using a port-shared address,
1034 blacklisting of their IP address will result in all other subscribers
1035 sharing that address having their ability to source SMTP packets
1036 restricted to some extent.
1037
1038 <span class="h3"><a class="selflink" id="section-13.4" href="#section-13.4">13.4</a>. Port Randomization</span>
1039
1040 A blind attack that can be performed against TCP relies on the
1041 attacker's ability to guess the 5-tuple (Protocol, Source Address,
1042 Destination Address, Source Port, Destination Port) that identifies
1043 the transport protocol instance to be attacked. [<a href="./rfc6056" title="&quot;Recommendations for Transport- Protocol Port Randomization&quot;">RFC6056</a>] describes
1044 a number of methods for the random selection of the source port
1045 number, such that the ability of an attacker to correctly guess the
1046 5-tuple is reduced. With shared IPv4 addresses, the port selection
1047 space is reduced. Preserving port randomization is important and may
1048 be more or less difficult depending on the address sharing solution
1049 and the size of the port space that is being manipulated. Allocation
1050 of non-contiguous port ranges could help to mitigate this issue.
1051
1052 It should be noted that guessing the port information may not be
1053 sufficient to carry out a successful blind attack. An in-window TCP
1054 Sequence Number (SN) should also be known or guessed. A TCP segment
1055 is processed only if all previous segments have been received, except
1056 for some Reset segment implementations that immediately process the
1057
1058
1059
1060 <span class="grey">Ford, et al. Informational [Page 19]</span></pre>
1061 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-20" ></span>
1062 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1063
1064
1065 Reset as long as it is within the Window. If SN is randomly chosen,
1066 it will be difficult to guess it (SN is 32 bits long); port
1067 randomization is one protection among others against blind attacks.
1068 There is more detailed discussion of improving TCP's robustness to
1069 Blind In-Window Attacks in [<a href="./rfc5961" title="&quot;Improving TCP's Robustness to Blind In-Window Attacks&quot;">RFC5961</a>].
1070
1071 <span class="h3"><a class="selflink" id="section-13.5" href="#section-13.5">13.5</a>. IPsec</span>
1072
1073 The impact of large-scale IP address sharing for IPsec operation
1074 should be evaluated and assessed. [<a href="./rfc3947" title="&quot;Negotiation of NAT-Traversal in the IKE&quot;">RFC3947</a>] proposes a solution to
1075 solve issues documented in [<a href="./rfc3715" title="&quot;IPsec-Network Address Translation (NAT) Compatibility Requirements&quot;">RFC3715</a>]. [<a href="./rfc5996" title="&quot;Internet Key Exchange Protocol Version 2 (IKEv2)&quot;">RFC5996</a>] specifies Internet
1076 Key Exchange (IKE) Protocol Version 2, which includes NAT traversal
1077 mechanisms that are now widely used to enable IPsec to work in the
1078 presence of NATs in many cases. Nevertheless, service providers may
1079 wish to ensure that CGN deployments do not inadvertently block NAT
1080 traversal for security protocols such as IKE (refer to [<a href="#ref-NAT-SEC" title="&quot;Security implications of Network Address Translators (NATs)&quot;">NAT-SEC</a>] for
1081 more information).
1082
1083 <span class="h3"><a class="selflink" id="section-13.6" href="#section-13.6">13.6</a>. Policing Forwarding Behavior</span>
1084
1085 [<a id="ref-RFC2827">RFC2827</a>] motivates and discusses a simple, effective, and
1086 straightforward method for using ingress traffic filtering to
1087 prohibit DoS attacks that use forged IP addresses. Following this
1088 recommendation, service providers operating shared-addressing
1089 mechanisms should ensure that source addresses, or source ports in
1090 the case of port-range schemes, are set correctly in outgoing packets
1091 from their subscribers or they should drop the packets.
1092
1093 If some form of IPv6 ingress filtering is deployed in the broadband
1094 network and DS-Lite service is restricted to those subscribers, then
1095 tunnels terminating at the CGN and coming from registered subscriber
1096 IPv6 addresses cannot be spoofed. Thus, a simple access control list
1097 on the tunnel transport source address is all that is required to
1098 accept traffic on the internal interface of a CGN.
1099
1100 <span class="h2"><a class="selflink" id="section-14" href="#section-14">14</a>. Transport Issues</span>
1101
1102 <span class="h3"><a class="selflink" id="section-14.1" href="#section-14.1">14.1</a>. Parallel Connections</span>
1103
1104 One issue is systems that assume that multiple simultaneous
1105 connections to a single IP address implies connectivity to a single
1106 host -- such systems may experience unexpected results.
1107
1108 <span class="h3"><a class="selflink" id="section-14.2" href="#section-14.2">14.2</a>. Serial Connections</span>
1109
1110 Another issue is systems that assume that returning to a given IP
1111 address means returning to the same physical host, as with stateful
1112 transactions. This may also affect cookie-based systems.
1113
1114
1115
1116 <span class="grey">Ford, et al. Informational [Page 20]</span></pre>
1117 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-21" ></span>
1118 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1119
1120
1121 <span class="h3"><a class="selflink" id="section-14.3" href="#section-14.3">14.3</a>. TCP Control Block Sharing</span>
1122
1123 [<a id="ref-RFC2140">RFC2140</a>] defines a performance optimization for TCP based on sharing
1124 state between TCP control blocks that pertain to connections to the
1125 same host, as opposed to maintaining state for each discrete
1126 connection. This optimization assumes that an address says something
1127 about the properties of the path between two hosts, which is clearly
1128 not the case if the address in question is shared by multiple hosts
1129 at different physical network locations. While CPE NAT today causes
1130 problems for sharing TCP control block state across multiple
1131 connections to a given IP address, large-scale address sharing will
1132 make these issues more severe and more widespread.
1133
1134 <span class="h2"><a class="selflink" id="section-15" href="#section-15">15</a>. Reverse DNS</span>
1135
1136 Many service providers populate forward and reverse DNS zones for the
1137 public IPv4 addresses that they allocate to their subscribers. In
1138 the case where public addresses are shared across multiple
1139 subscribers, such strings are, by definition, no longer sufficient to
1140 identify an individual subscriber without additional information.
1141
1142 <span class="h2"><a class="selflink" id="section-16" href="#section-16">16</a>. Load Balancing</span>
1143
1144 Algorithms used to balance traffic load for popular destinations may
1145 be affected by the introduction of address sharing. Where balancing
1146 is achieved by deterministically routing traffic from specific source
1147 IP addresses to specific servers, imbalances in load may be
1148 experienced as address sharing is enabled for some of those source IP
1149 addresses. This will require re-evaluation of the algorithms used in
1150 the load-balancing design. In general, as the scale of address
1151 sharing grows, load-balancing designs will need to be re-evaluated
1152 and any assumptions about average load per source IP address
1153 revisited.
1154
1155 <span class="h2"><a class="selflink" id="section-17" href="#section-17">17</a>. IPv6 Transition Issues</span>
1156
1157 IPv4 address sharing solutions may interfere with existing IPv4 to
1158 IPv6 transition mechanisms, which were not designed with IPv4
1159 shortage considerations in mind. With port-range solutions, for
1160 instance, incoming 6to4 packets should be able to find their way from
1161 a 6to4 relay to the appropriate 6to4 CPE router, despite the lack of
1162 direct port-range information (UDP/TCP initial source port did not
1163 pass through the CPE port range translation process). One solution
1164 would be for a 6to4 IPv6 address to embed not only an IPv4 address
1165 but also a port range value.
1166
1167
1168
1169
1170
1171
1172 <span class="grey">Ford, et al. Informational [Page 21]</span></pre>
1173 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-22" ></span>
1174 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1175
1176
1177 Subscribers allocated with private addresses will not be able to
1178 utilize 6to4 [<a href="./rfc3056" title="&quot;Connection of IPv6 Domains via IPv4 Clouds&quot;">RFC3056</a>] to access IPv6, but may be able to utilize
1179 Teredo [<a href="./rfc4380" title="&quot;Teredo: Tunneling IPv6 over UDP through Network Address Translations (NATs)&quot;">RFC4380</a>].
1180
1181 Some routers enable 6to4 on their WAN link. 6to4 requires a publicly
1182 routable IPv4 address. Enabling 6to4 when the apparently public IPv4
1183 WAN address is in fact behind a NAT creates a disconnected IPv6
1184 island.
1185
1186 <span class="h2"><a class="selflink" id="section-18" href="#section-18">18</a>. Introduction of Single Points of Failure</span>
1187
1188 In common with all deployments of new network functionality, the
1189 introduction of new nodes or functions to handle the multiplexing of
1190 multiple subscribers across shared IPv4 addresses could create single
1191 points of failure in the network. Any IP address sharing solution
1192 should consider the opportunity to add redundancy features in order
1193 to alleviate the impact on the robustness of the offered IP
1194 connectivity service. The ability of the solution to allow hot
1195 swapping from one machine to another should be considered. This is
1196 especially important where the address sharing solution in question
1197 requires the creation of per-flow state in the network.
1198
1199 <span class="h2"><a class="selflink" id="section-19" href="#section-19">19</a>. State Maintenance Reduces Battery Life</span>
1200
1201 In order for hosts to maintain network state in the presence of NAT,
1202 keep-alive messages have to be sent at frequent intervals. For
1203 battery-powered devices, sending these keep-alive messages can result
1204 in significantly reduced battery performance than would otherwise be
1205 the case [<a href="#ref-Mobile_Energy_Consumption">Mobile_Energy_Consumption</a>].
1206
1207 <span class="h2"><a class="selflink" id="section-20" href="#section-20">20</a>. Support of Multicast</span>
1208
1209 [<a id="ref-RFC5135">RFC5135</a>] specifies requirements for a NAT that supports Any Source
1210 IP Multicast or Source-Specific IP Multicast. Port-range routers
1211 that form part of port-range solutions will need to support similar
1212 requirements if multicast support is required.
1213
1214 <span class="h2"><a class="selflink" id="section-21" href="#section-21">21</a>. Support of Mobile-IP</span>
1215
1216 IP address sharing within the context of Mobile IP deployments (in
1217 the home network and/or in the visited network) will require Home
1218 Agents and/or Foreign Agents to be updated so as to take into account
1219 the relevant port information. There may also be issues raised when
1220 an additional layer of encapsulation is required thereby causing, or
1221 increasing the need for, fragmentation and reassembly.
1222
1223 Issues for Mobile-IP in the presence of NAT are discussed in
1224 [<a href="#ref-NAT64-MOBILITY">NAT64-MOBILITY</a>].
1225
1226
1227
1228 <span class="grey">Ford, et al. Informational [Page 22]</span></pre>
1229 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-23" ></span>
1230 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1231
1232
1233 <span class="h2"><a class="selflink" id="section-22" href="#section-22">22</a>. Security Considerations</span>
1234
1235 This memo does not define any protocol and therefore creates no new
1236 security issues. <a href="#section-13">Section 13</a> discusses some of the security and
1237 identity-related implications of IP address sharing.
1238
1239 <span class="h2"><a class="selflink" id="section-23" href="#section-23">23</a>. Contributors</span>
1240
1241 This document is based on sources co-authored by J.L. Grimault and A.
1242 Villefranque of France Telecom.
1243
1244 <span class="h2"><a class="selflink" id="section-24" href="#section-24">24</a>. Acknowledgments</span>
1245
1246 This memo was partly inspired by conversations that took place as
1247 part of Internet Society (ISOC) hosted roundtable events for
1248 operators and content providers deploying IPv6. Participants in
1249 those discussions included John Brzozowski, Leslie Daigle, Tom
1250 Klieber, Yiu Lee, Kurtis Lindqvist, Wes George, Lorenzo Colliti, Erik
1251 Kline, Igor Gashinsky, Jason Fesler, Rick Reed, Adam Bechtel, Larry
1252 Campbell, Tom Coffeen, David Temkin, Pete Gelbman, Mark Winter, Will
1253 Charnock, Martin Levy, Greg Wood, and Christian Jacquenet.
1254
1255 The authors are also grateful to Christian Jacquenet, Iain Calder,
1256 Joel Halpern, Brian Carpenter, Gregory Lebovitz, Bob Briscoe, Marcelo
1257 Bagnulo, Dan Wing, and Wes George for their helpful comments and
1258 suggestions for improving the document.
1259
1260 This memo was created using the xml2rfc tool.
1261
1262 <span class="h2"><a class="selflink" id="section-25" href="#section-25">25</a>. Informative References</span>
1263
1264 [A+P] Bush, R., "The A+P Approach to the IPv4 Address Shortage",
1265 Work in Progress, February 2011.
1266
1267 [<a id="ref-CGN_Viability">CGN_Viability</a>]
1268 Alcock, S., "Research into the Viability of Service-
1269 Provider NAT", 2008, &lt;<a href="http://www.wand.net.nz/~salcock/someisp/flow_counting/result_page.html">http://www.wand.net.nz/~salcock/</a>
1270 <a href="http://www.wand.net.nz/~salcock/someisp/flow_counting/result_page.html">someisp/flow_counting/result_page.html</a>&gt;.
1271
1272 [<a id="ref-DS-Lite">DS-Lite</a>] Durand, A., Droms, R., Woodyatt, J., and Y. Lee, "Dual-
1273 Stack Lite Broadband Deployments Following IPv4
1274 Exhaustion", Work in Progress, May 2011.
1275
1276 [<a id="ref-IANA_Ports">IANA_Ports</a>]
1277 IANA, "Port Numbers", &lt;<a href="http://www.iana.org">http://www.iana.org</a>&gt;.
1278
1279
1280
1281
1282
1283
1284 <span class="grey">Ford, et al. Informational [Page 23]</span></pre>
1285 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-24" ></span>
1286 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1287
1288
1289 [<a id="ref-IPv4_Pool">IPv4_Pool</a>]
1290 ICANN, "Available Pool of Unallocated IPv4 Internet
1291 Addresses Now Completely Emptied", February 2011,
1292 &lt;<a href="http://icann.org/en/news/releases/release-03feb11-en.pdf">http://icann.org/en/news/releases/</a>
1293 <a href="http://icann.org/en/news/releases/release-03feb11-en.pdf">release-03feb11-en.pdf</a>&gt;.
1294
1295 [<a id="ref-LSN-REQS">LSN-REQS</a>] Perreault, S., Yamagata, I., Miyakawa, S., Nakagawa, A.,
1296 and H. Ashida, "Common requirements for IP address sharing
1297 schemes", Work in Progress, March 2011.
1298
1299 [<a id="ref-Mobile_Energy_Consumption">Mobile_Energy_Consumption</a>]
1300 Haverinen, H., Siren, J., and P. Eronen, "Energy
1301 Consumption of Always-On Applications in WCDMA Networks",
1302 April 2007, &lt;<a href="http://research.nokia.com/node/5597">http://research.nokia.com/node/5597</a>&gt;.
1303
1304 [<a id="ref-NAT-PMP">NAT-PMP</a>] Cheshire, S., <a style="text-decoration: none" href='https://www.google.com/search?sitesearch=datatracker.ietf.org%2Fdoc%2Fhtml%2F&amp;q=inurl:draft-+%22NAT+Port+Mapping+Protocol+%28NAT-PMP%29%22'>"NAT Port Mapping Protocol (NAT-PMP)"</a>, Work
1305 in Progress, April 2008.
1306
1307 [<a id="ref-NAT-SEC">NAT-SEC</a>] Gont, F. and P. Srisuresh, "Security implications of
1308 Network Address Translators (NATs)", Work in Progress,
1309 October 2009.
1310
1311 [<a id="ref-NAT444">NAT444</a>] Yamagata, I., Shirasaki, Y., Nakagawa, A., Yamaguchi, J.,
1312 and H. Ashida, "NAT444", January 2011.
1313
1314 [<a id="ref-NAT64-MOBILITY">NAT64-MOBILITY</a>]
1315 Haddad, W. and C. Perkins, "A Note on NAT64 Interaction
1316 with Mobile IPv6", Work in Progress, March 2011.
1317
1318 [<a id="ref-PCP">PCP</a>] Wing, D., Ed., Cheshire, S., Boucadair, M., Penno, R., and
1319 P. Selkirk, "Port Control Protocol (PCP)", Work
1320 in Progress, May 2011.
1321
1322 [<a id="ref-PORT-RANGE">PORT-RANGE</a>]
1323 Boucadair, M., Levis, P., Bajko, G., and T. Savolainen,
1324 "IPv4 Connectivity Access in the Context of IPv4 Address
1325 Exhaustion: Port Range based IP Architecture", Work
1326 in Progress, July 2009.
1327
1328 [<a id="ref-RFC1191">RFC1191</a>] Mogul, J. and S. Deering, "Path MTU discovery", <a href="./rfc1191">RFC 1191</a>,
1329 November 1990.
1330
1331 [<a id="ref-RFC1337">RFC1337</a>] Braden, B., "TIME-WAIT Assassination Hazards in TCP",
1332 <a href="./rfc1337">RFC 1337</a>, May 1992.
1333
1334 [<a id="ref-RFC1700">RFC1700</a>] Reynolds, J. and J. Postel, "Assigned Numbers", <a href="./rfc1700">RFC 1700</a>,
1335 October 1994.
1336
1337
1338
1339
1340 <span class="grey">Ford, et al. Informational [Page 24]</span></pre>
1341 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-25" ></span>
1342 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1343
1344
1345 [<a id="ref-RFC1918">RFC1918</a>] Rekhter, Y., Moskowitz, R., Karrenberg, D., Groot, G., and
1346 E. Lear, "Address Allocation for Private Internets",
1347 <a href="https://www.rfc-editor.org/bcp/bcp5">BCP 5</a>, <a href="./rfc1918">RFC 1918</a>, February 1996.
1348
1349 [<a id="ref-RFC2140">RFC2140</a>] Touch, J., "TCP Control Block Interdependence", <a href="./rfc2140">RFC 2140</a>,
1350 April 1997.
1351
1352 [<a id="ref-RFC2663">RFC2663</a>] Srisuresh, P. and M. Holdrege, "IP Network Address
1353 Translator (NAT) Terminology and Considerations",
1354 <a href="./rfc2663">RFC 2663</a>, August 1999.
1355
1356 [<a id="ref-RFC2782">RFC2782</a>] Gulbrandsen, A., Vixie, P., and L. Esibov, "A DNS RR for
1357 specifying the location of services (DNS SRV)", <a href="./rfc2782">RFC 2782</a>,
1358 February 2000.
1359
1360 [<a id="ref-RFC2827">RFC2827</a>] Ferguson, P. and D. Senie, "Network Ingress Filtering:
1361 Defeating Denial of Service Attacks which employ IP Source
1362 Address Spoofing", <a href="https://www.rfc-editor.org/bcp/bcp38">BCP 38</a>, <a href="./rfc2827">RFC 2827</a>, May 2000.
1363
1364 [<a id="ref-RFC2993">RFC2993</a>] Hain, T., "Architectural Implications of NAT", <a href="./rfc2993">RFC 2993</a>,
1365 November 2000.
1366
1367 [<a id="ref-RFC3056">RFC3056</a>] Carpenter, B. and K. Moore, "Connection of IPv6 Domains
1368 via IPv4 Clouds", <a href="./rfc3056">RFC 3056</a>, February 2001.
1369
1370 [<a id="ref-RFC3232">RFC3232</a>] Reynolds, J., "Assigned Numbers: <a href="./rfc1700">RFC 1700</a> is Replaced by
1371 an On-line Database", <a href="./rfc3232">RFC 3232</a>, January 2002.
1372
1373 [<a id="ref-RFC3715">RFC3715</a>] Aboba, B. and W. Dixon, "IPsec-Network Address Translation
1374 (NAT) Compatibility Requirements", <a href="./rfc3715">RFC 3715</a>, March 2004.
1375
1376 [<a id="ref-RFC3947">RFC3947</a>] Kivinen, T., Swander, B., Huttunen, A., and V. Volpe,
1377 "Negotiation of NAT-Traversal in the IKE", <a href="./rfc3947">RFC 3947</a>,
1378 January 2005.
1379
1380 [<a id="ref-RFC4380">RFC4380</a>] Huitema, C., "Teredo: Tunneling IPv6 over UDP through
1381 Network Address Translations (NATs)", <a href="./rfc4380">RFC 4380</a>,
1382 February 2006.
1383
1384 [<a id="ref-RFC4787">RFC4787</a>] Audet, F. and C. Jennings, "Network Address Translation
1385 (NAT) Behavioral Requirements for Unicast UDP", <a href="https://www.rfc-editor.org/bcp/bcp127">BCP 127</a>,
1386 <a href="./rfc4787">RFC 4787</a>, January 2007.
1387
1388 [<a id="ref-RFC5135">RFC5135</a>] Wing, D. and T. Eckert, "IP Multicast Requirements for a
1389 Network Address Translator (NAT) and a Network Address
1390 Port Translator (NAPT)", <a href="https://www.rfc-editor.org/bcp/bcp135">BCP 135</a>, <a href="./rfc5135">RFC 5135</a>, February 2008.
1391
1392
1393
1394
1395
1396 <span class="grey">Ford, et al. Informational [Page 25]</span></pre>
1397 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-26" ></span>
1398 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1399
1400
1401 [<a id="ref-RFC5508">RFC5508</a>] Srisuresh, P., Ford, B., Sivakumar, S., and S. Guha, "NAT
1402 Behavioral Requirements for ICMP", <a href="https://www.rfc-editor.org/bcp/bcp148">BCP 148</a>, <a href="./rfc5508">RFC 5508</a>,
1403 April 2009.
1404
1405 [<a id="ref-RFC5961">RFC5961</a>] Ramaiah, A., Stewart, R., and M. Dalal, "Improving TCP's
1406 Robustness to Blind In-Window Attacks", <a href="./rfc5961">RFC 5961</a>,
1407 August 2010.
1408
1409 [<a id="ref-RFC5996">RFC5996</a>] Kaufman, C., Hoffman, P., Nir, Y., and P. Eronen,
1410 "Internet Key Exchange Protocol Version 2 (IKEv2)",
1411 <a href="./rfc5996">RFC 5996</a>, September 2010.
1412
1413 [<a id="ref-RFC6056">RFC6056</a>] Larsen, M. and F. Gont, "Recommendations for Transport-
1414 Protocol Port Randomization", <a href="https://www.rfc-editor.org/bcp/bcp156">BCP 156</a>, <a href="./rfc6056">RFC 6056</a>,
1415 January 2011.
1416
1417 [<a id="ref-RFC6145">RFC6145</a>] Li, X., Bao, C., and F. Baker, "IP/ICMP Translation
1418 Algorithm", <a href="./rfc6145">RFC 6145</a>, April 2011.
1419
1420 [<a id="ref-RFC6146">RFC6146</a>] Bagnulo, M., Matthews, P., and I. van Beijnum, "Stateful
1421 NAT64: Network Address and Protocol Translation from IPv6
1422 Clients to IPv4 Servers", <a href="./rfc6146">RFC 6146</a>, April 2011.
1423
1424 [<a id="ref-SAM">SAM</a>] Despres, R., "Scalable Multihoming across IPv6 Local-
1425 Address Routing Zones Global-Prefix/Local-Address
1426 Stateless Address Mapping (SAM)", July 2009.
1427
1428 [<a id="ref-UPnP-IGD">UPnP-IGD</a>] UPnP Forum, "Universal Plug and Play (UPnP) Internet
1429 Gateway Device (IGD) V 2.0", December 2010,
1430 &lt;<a href="http://upnp.org/specs/gw/igd2/">http://upnp.org/specs/gw/igd2/</a>&gt;.
1431
1432 [<a id="ref-Windows-Logo">Windows-Logo</a>]
1433 Microsoft, "Windows Logo Program Requirements and
1434 Policies", 2006, &lt;<a href="http://www.microsoft.com/whdc/winlogo/hwrequirements/default.mspx">http://www.microsoft.com/whdc/winlogo/</a>
1435 <a href="http://www.microsoft.com/whdc/winlogo/hwrequirements/default.mspx">hwrequirements/default.mspx</a>&gt;.
1436
1437
1438
1439
1440
1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452 <span class="grey">Ford, et al. Informational [Page 26]</span></pre>
1453 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-27" ></span>
1454 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1455
1456
1457 <span class="h2"><a class="selflink" id="appendix-A" href="#appendix-A">Appendix A</a>. Classes of Address Sharing Solution</span>
1458
1459 IP address sharing solutions fall into two classes. Either a
1460 service-provider-operated NAT function is introduced and subscribers
1461 are allocated addresses from [<a href="./rfc1918" title="&quot;Address Allocation for Private Internets&quot;">RFC1918</a>] space, or public IPv4
1462 addresses are shared across multiple subscribers by restricting the
1463 range of ports available to each subscriber. These classes of
1464 solution are described in a bit more detail below.
1465
1466 o CGN-based solutions: These solutions propose the introduction of a
1467 NAPT function in the service provider's network, denoted also as
1468 Carrier Grade NAT (CGN), or Large Scale NAT (LSN) [<a href="#ref-LSN-REQS" title="&quot;Common requirements for IP address sharing schemes&quot;">LSN-REQS</a>], or
1469 Provider NAT. The CGN is responsible for translating private
1470 addresses to publicly routable addresses. Private addresses are
1471 assigned to subscribers, a pool of public addresses is assigned to
1472 the CGN, and the number of public addresses is smaller than the
1473 number of subscribers. A public IPv4 address in the CGN pool is
1474 shared by several subscribers at the same time. Solutions making
1475 use of a service provider-based NAT include [<a href="#ref-NAT444" title="&quot;NAT444&quot;">NAT444</a>] (two layers
1476 of NAT) and [<a href="#ref-DS-Lite" title="&quot;Dual- Stack Lite Broadband Deployments Following IPv4 Exhaustion&quot;">DS-Lite</a>] (a single layer of NAT).
1477
1478 o Port-range solutions: These solutions avoid the presence of a CGN
1479 function. A single public IPv4 address is assigned to several
1480 subscribers at the same time. A restricted port range is also
1481 assigned to each subscriber so that two subscribers with the same
1482 IPv4 address have two different port ranges that do not overlap.
1483 These solutions are called Address+Port [A+P], or Port Range
1484 [<a href="#ref-PORT-RANGE">PORT-RANGE</a>], or Stateless Address Mapping [<a href="#ref-SAM" title="&quot;Scalable Multihoming across IPv6 Local- Address Routing Zones Global-Prefix/Local-Address Stateless Address Mapping (SAM)&quot;">SAM</a>].
1485
1486 <span class="h2"><a class="selflink" id="appendix-B" href="#appendix-B">Appendix B</a>. Address Space Multiplicative Factor</span>
1487
1488 The purpose of sharing public IPv4 addresses is to increase the
1489 addressing space. A key parameter is the factor by which service
1490 providers want or need to multiply their IPv4 public address space,
1491 and the consequence is the number of subscribers sharing the same
1492 public IPv4 address. We refer to this parameter as the address space
1493 multiplicative factor; the inverse is called the compression ratio.
1494
1495 The multiplicative factor can only be applied to the subset of
1496 subscribers that are eligible for a shared address. The reasons a
1497 subscriber cannot have a shared address can be:
1498
1499 o It would not be compatible with the service to which they are
1500 currently subscribed (for example, business subscriber).
1501
1502
1503
1504
1505
1506
1507
1508 <span class="grey">Ford, et al. Informational [Page 27]</span></pre>
1509 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-28" ></span>
1510 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1511
1512
1513 o Subscriber CPE is not compatible with the address sharing solution
1514 selected by the service provider (for example, it does not handle
1515 port restriction for port-range solutions or it does not allow
1516 IPv4 in IPv6 encapsulation for the DS-Lite solution), and its
1517 replacement is not easy.
1518
1519 Different service providers may have very different needs. A long-
1520 lived service provider, whose number of subscribers is rather stable,
1521 may have an existing address pool that will only need a small
1522 extension to cope with the next few years, assuming that this address
1523 pool can be re-purposed for an address sharing solution (small
1524 multiplicative factor, less than 10). A new entrant or a new line of
1525 business will need a much bigger multiplicative factor (e.g., 1000).
1526 A mobile operator may see its addressing needs grow dramatically as
1527 the IP-enabled mobile handset market grows.
1528
1529 When the multiplicative factor is large, the average number of ports
1530 per subscriber is small. Given the large measured disparity between
1531 average and peak port consumption [<a href="#ref-CGN_Viability">CGN_Viability</a>], this will create
1532 service problems in the event that ports are allocated statically.
1533 In this case, it is essential for port allocation to map to need as
1534 closely as possible, and to avoid allocating ports for longer than
1535 necessary. Therefore, the larger the multiplicative factor, the more
1536 dynamic the port assignment has to be.
1537
1538
1539
1540
1541
1542
1543
1544
1545
1546
1547
1548
1549
1550
1551
1552
1553
1554
1555
1556
1557
1558
1559
1560
1561
1562
1563
1564 <span class="grey">Ford, et al. Informational [Page 28]</span></pre>
1565 <hr class='noprint'/><!--NewPage--><pre class='newpage'><span id="page-29" ></span>
1566 <span class="grey"><a href="./rfc6269">RFC 6269</a> Issues with IP Address Sharing June 2011</span>
1567
1568
1569 Authors' Addresses
1570
1571 Mat Ford (editor)
1572 Internet Society
1573 Geneva
1574 Switzerland
1575
1576 EMail: [email protected]
1577
1578
1579 Mohamed Boucadair
1580 France Telecom
1581 Rennes 35000
1582 France
1583
1584 EMail: [email protected]
1585
1586
1587 Alain Durand
1588 Juniper Networks
1589
1590 EMail: [email protected]
1591
1592
1593 Pierre Levis
1594 France Telecom
1595 42 rue des Coutures
1596 BP 6243
1597 Caen Cedex 4 14066
1598 France
1599
1600 EMail: [email protected]
1601
1602
1603 Phil Roberts
1604 Internet Society
1605 Reston, VA
1606 USA
1607
1608 EMail: [email protected]
1609
1610
1611
1612
1613
1614
1615
1616
1617
1618
1619
1620 Ford, et al. Informational [Page 29]
1621 </pre>