idnits 2.17.1 draft-ietf-lemonade-deployments-08.txt: Checking boilerplate required by RFC 5378 and the IETF Trust (see https://trustee.ietf.org/license-info): ---------------------------------------------------------------------------- ** It looks like you're using RFC 3978 boilerplate. You should update this to the boilerplate described in the IETF Trust License Policy document (see https://trustee.ietf.org/license-info), which is required now. -- Found old boilerplate from RFC 3978, Section 5.1 on line 13. -- Found old boilerplate from RFC 3978, Section 5.5, updated by RFC 4748 on line 597. -- Found old boilerplate from RFC 3979, Section 5, paragraph 1 on line 566. -- Found old boilerplate from RFC 3979, Section 5, paragraph 2 on line 573. -- Found old boilerplate from RFC 3979, Section 5, paragraph 3 on line 580. Checking nits according to https://www.ietf.org/id-info/1id-guidelines.txt: ---------------------------------------------------------------------------- == No 'Intended status' indicated for this document; assuming Proposed Standard Checking nits according to https://www.ietf.org/id-info/checklist : ---------------------------------------------------------------------------- No issues found here. Miscellaneous warnings: ---------------------------------------------------------------------------- == The copyright year in the IETF Trust Copyright Line does not match the current year -- The document seems to lack a disclaimer for pre-RFC5378 work, but may have content which was first submitted before 10 November 2008. If you have contacted all the original authors and they are all willing to grant the BCP78 rights to the IETF Trust, then this is fine, and you can ignore this comment. If not, you may need to add the pre-RFC5378 disclaimer. (See the Legal Provisions document at https://trustee.ietf.org/license-info for more information.) -- The document date (May 2007) is 6191 days in the past. Is this intentional? Checking references for intended status: Proposed Standard ---------------------------------------------------------------------------- (See RFCs 3967 and 4897 for information about using normative references to lower-maturity documents in RFCs) -- Possible downref: Non-RFC (?) normative reference: ref. 'IANA' ** Obsolete normative reference: RFC 793 (ref. 'TCP') (Obsoleted by RFC 9293) -- Obsolete informational reference (is this intentional?): RFC 3501 (ref. 'IMAP') (Obsoleted by RFC 9051) -- Obsolete informational reference (is this intentional?): RFC 4550 (ref. 'PROFILE') (Obsoleted by RFC 5550) -- Obsolete informational reference (is this intentional?): RFC 4409 (ref. 'SUBMISSION') (Obsoleted by RFC 6409) Summary: 2 errors (**), 0 flaws (~~), 2 warnings (==), 11 comments (--). Run idnits with the --verbose option for more detailed information about the items above. -------------------------------------------------------------------------------- 1 Internet Draft: Deployment Considerations for 2 lemonade-compliant Mobile Email R. Gellens 3 Document: draft-ietf-lemonade-deployments-08.txt Qualcomm 4 Expires: November 2007 May 2007 6 Deployment Considerations for lemonade-compliant Mobile Email 8 Status of this Memo 10 By submitting this Internet-Draft, each author represents that any 11 applicable patent or other IPR claims of which he or she is aware 12 have been or will be disclosed, and any of which he or she becomes 13 aware will be disclosed, in accordance with Section 6 of BCP 79. 15 Internet-Drafts are working documents of the Internet Engineering 16 Task Force (IETF), its areas, and its working groups. Note that 17 other groups may also distribute working documents as Internet- 18 Drafts. 20 Internet-Drafts are draft documents valid for a maximum of six 21 months and may be updated, replaced, or obsoleted by other documents 22 at any time. It is inappropriate to use Internet-Drafts as 23 reference material or to cite them other than as "work in progress." 25 The list of current Internet-Drafts can be accessed at 26 http://www.ietf.org/ietf/1id-abstracts.txt The list of 27 Internet-Draft Shadow Directories can be accessed at 28 http://www.ietf.org/shadow.html. 30 Copyright Notice 32 Copyright (C) The IETF Trust (2007). All Rights Reserved. 34 Abstract 36 This document discusses deployment issues and describes requirements 37 for successful deployment of mobile email which are implicit in the 38 IETF lemonade documents. 40 Gellens [Page 1] Expires November 2007 41 Table of Contents 43 1 Conventions Used in this Document . . . . . . . . . . . . . . 2 44 2 Introduction . . . . . . . . . . . . . . . . . . . . . . . . 2 45 3 Ports . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 46 4 TCP Connections . . . . . . . . . . . . . . . . . . . . . . 3 47 4.1 Lifetime . . . . . . . . . . . . . . . . . . . . . . . . 4 48 4.2 Maintenance during temporary transport loss . . . . . . 5 49 5 Dormancy . . . . . . . . . . . . . . . . . . . . . . . . . . 5 50 6 Firewalls . . . . . . . . . . . . . . . . . . . . . . . . . 6 51 6.1 Firewall Traversal . . . . . . . . . . . . . . . . . . . 7 52 7 NATs . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 53 8 Security Considerations . . . . . . . . . . . . . . . . . . . 8 54 9 IANA Considerations . . . . . . . . . . . . . . . . . . . . 10 55 10 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . 10 56 11 Normative References . . . . . . . . . . . . . . . . . . . . 10 57 12 Informative References . . . . . . . . . . . . . . . . . . . 10 58 13 Author's Address . . . . . . . . . . . . . . . . . . . . . . 11 59 Appendix A: Changes from Previous Version . . . . . . . . . . 11 60 Intellectual Property Statement . . . . . . . . . . . . . . . 12 61 Full Copyright Statement . . . . . . . . . . . . . . . . . . . 13 63 1 Conventions Used in this Document 65 The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT", 66 "SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this 67 document are to be interpreted as described in [KEYWORDS]. 69 2 Introduction 71 The IETF lemonade group has developed a set of extensions to IMAP 72 and Message Submission, along with a profile document which 73 restricts server behavior and describes client usage [PROFILE]. 75 Successful deployment of lemonade-compliant mobile email requires 76 various functionality which is generally assumed and hence often not 77 covered in email RFCs. This document describes some of these 78 additional considerations, with a focus on those which have been 79 reported to be problematic. 81 3 Ports 83 Both IMAP and Message Submission have been assigned well-known ports 84 [IANA] which MUST be available. IMAP uses port 143. Message 85 Submission uses port 587. It is REQUIRED that the client be able to 86 contact the server on these ports. Hence the client and server 88 Gellens [Page 2] Expires November 2007 89 systems, as well as any intermediary systems, MUST allow 90 communication on these ports. 92 Historically, MUAs have used port 25 for message submission, and 93 [SUBMISSION] does accommodate this. However, it has become 94 increasingly common for ISPs and organizations to restrict outbound 95 port 25. Additionally, hotels and other public accommodations 96 sometimes intercept port 25 connections, regardless of the 97 destination host, resulting in users unexpectedly submitting 98 potentially sensitive communications to unknown and untrusted 99 third-party servers. Typically, users are not aware of such 100 interception. (Such interception violates [FIREWALLS] and has many 101 negative consequences.) 103 Due to endemic security vulnerabilities in widely-deployed SMTP 104 servers, organizations often employ application-level firewalls 105 which intercept SMTP and permit only a limited subset of the 106 protocol. New extensions are therefore more difficult to deploy on 107 port 25. Since lemonade requires support for several [SUBMISSION] 108 extensions, it is extremely important that lemonade clients use, and 109 lemonade servers listen on, port 587 by default. 111 In addition to communications between the client and server systems, 112 lemonade requires that the Message Submission server be able to 113 establish a TCP connection to the IMAP server (for 114 forward-without-download). This uses port 143 by default. 116 Messaging clients sometimes use protocols to store, retrieve, and 117 update configuration and preference data. Functionality such as 118 setting a new device to use the configuration and preference data of 119 another device, or having a device inherit default configuration 120 data from a user account, an organization, or other source, is 121 likely to be even more useful with small mobile devices. One such 122 protocol which was developed for this purpose is [ACAP]. It is 123 therefore RECOMMENDED that clients be able to contact servers on 124 this port (674). 126 Note that systems which do not support application use of [TCP] on 127 arbitrary ports are not full Internet clients. As a result, such 128 systems use gateways to the Internet which necessarily result in 129 data integrity problems. 131 4 TCP Connections 133 Both IMAP and Message Submission use [TCP]. Hence the client system 134 MUST be able to establish and maintain TCP connections to these 135 servers. The Message Submission server MUST be able to initiate a 136 connection to the IMAP server. Support for application use of [TCP] 138 Gellens [Page 3] Expires November 2007 139 is REQUIRED of both client and server systems. 141 The requirements and advice in [HOST-REQUIREMENTS] SHOULD be 142 followed. 144 Note that, for environments that do not support application use of 145 [TCP] but do so for HTTP, email can be offered by deploying webmail. 146 Webmail is a common term for email over the web, where a server 147 speaks HTTP to the client and an email protocol (often IMAP) to the 148 mail store. Its functionality is necessarily limited by the 149 capabilities of the web client, the webmail server, the protocols 150 used between the webmail server and the client (HTTP and a markup 151 language such as HTML), and between the webmail server and the mail 152 store. However, if HTTP is all that is available to an application, 153 the environment is by definition limited and thus functionality 154 offered to the user must also be limited, and can't be lemonade 155 compliant. 157 4.1 Lifetime 159 The duration of the TCP connections between the client and server 160 systems for both IMAP and Message Submission can be arbitrarily 161 long. The client system, the server, as well as all intermediate 162 systems MUST NOT terminate these TCP connections simply because of 163 their duration (that is, just because of how long they have been 164 open). 166 Lemonade depends on idle timers being enforced only at the 167 application level (IMAP and Message Submission): if no data is 168 received within a period of time, either side MAY terminate the 169 connection as permitted by the protocol (see [SUBMISSION] or 170 [IMAP]). Since IMAP permits unsolicited notifications of state 171 changes, it is reasonable for clients to remain connected for 172 extended periods with no data being exchanged. Being forced to send 173 data just to keep the connection alive can prevent or hinder 174 optimizations such as dormancy mode (see section 5). 176 Two hours is a fairly common configuration timeout at middleboxes. 177 That is, there are a number of sites at which TCP connections are 178 torn down by the network two hours after data was last sent in 179 either direction. Thus, lemonade clients and servers SHOULD make 180 sure that, in the absence of specific configuration otherwise, the 181 TCP connection not remain idle for two hours. This rule ensures 182 that, by default, lemonade clients and servers operate in 183 environments configured with a two-hour maximum for idle TCP 184 connections. Network and server operators can still permit IMAP 185 connections to remain idle in excess of two hours and thus increase 186 the benefits of dormancy, by configuring lemonade clients and 188 Gellens [Page 4] Expires November 2007 189 servers, and network equipment, to allow this. 191 It has been reported that some networks impose duration time 192 restrictions of their own on TCP connections other than HTTP. Such 193 behavior is harmful to email and all other TCP-based protocols. It 194 is unclear how widespread such reported behavior is, or if it is an 195 accidental consequence of an attempt at optimizing for HTTP traffic, 196 implementation limitations in firewalls, NATs or other devices, or a 197 deliberate choice. Either way, such a barrier to TCP connections is 198 a significant risk to the increasing usage of IETF protocols on such 199 networks. Note that TCP is designed to be more efficient when it is 200 used to transfer data over time. Prohibiting such connections thus 201 imposes hidden costs on an operator's network, forcing clients to 202 use TCP in inefficient ways. One way in which carriers can 203 inadvertently force TCP connections closed, resulting in users 204 wasting packets by reopening them, is described in Section 7 206 Note that systems remain able to terminate TCP connections at any 207 time based on local decisions, for example, to prevent overload 208 during a denial-of-service attack. These mechanisms are permitted 209 to take idle time into consideration and are not affected by these 210 requirements. 212 4.2 Maintenance during temporary transport loss 214 TCP is designed to withstand temporary loss of lower-level 215 connectivity. Such transient loss is not uncommon in mobile systems 216 (for example, due to handoffs, fade, etc.). The TCP connection 217 SHOULD be able to survive temporary lower-level loss when the IP 218 address of the client does not change (for example, short-duration 219 loss of the mobile device's traffic channel or periods of high 220 packet loss). Thus, the TCP/IP stack on the client, the server, and 221 all intermediate systems SHOULD maintain the TCP connection during 222 transient loss of connectivity. 224 To this end, client and server systems SHOULD NOT set the TCP 225 keep-alive socket option, and MUST NOT close a connection based on 226 ICMP "soft" errors, such as host unreachable messages. 228 5 Dormancy 230 Cellular data channels are connection-oriented (they are brought up 231 or down to establish or tear down connections); it costs network 232 resources to establish connections. Generally speaking, mobile 233 device battery charges last longer when the traffic channel is used 234 less. 236 Gellens [Page 5] Expires November 2007 237 Some mobile devices and networks support dormant mode, in which the 238 traffic channel is brought down during idle periods, yet the PPP or 239 equivalent level remains active, and the mobile retains its IP 240 address. 242 Maintenance of TCP connections during dormancy SHOULD be supported 243 by the client, server, and any intermediate systems. Thus, as 244 stated in 4.2 above, client and server systems SHOULD NOT set the 245 TCP keep-alive socket option, and MUST NOT close a connection based 246 on ICMP host unreachable messages. 248 Sending packets just to keep the session active causes unnecessary 249 channel establishment and timeout; with a long-idle TCP connection, 250 this would periodically bring up the channel and then let it idle 251 until it times out, again and again. 253 6 Firewalls 255 New services must necessarily have their traffic pass through 256 firewalls in order to be usable by corporate employees or 257 organization members connecting externally, such as when using 258 mobile devices. Firewalls exist to block traffic, yet exceptions 259 must be made for services to be used. There is a body of best 260 practices based on long experience in this area. Numerous 261 techniques exist to help organizations balance protecting themselves 262 and providing services to their members, employees, and/or 263 customers. (Describing, or even enumerating, such techniques and 264 practices is beyond the scope of this document, but section 8 does 265 mention some.) 267 It is critical that protocol design and architecture permit such 268 practices, and not constrain them. One key way in which the design 269 of a new service can aid its secure deployment is to maintain the 270 one-to-one association of services and port numbers. 272 One or more firewalls might exist in the path between the client and 273 server systems, as well as between the Message Submission and IMAP 274 servers. Proper deployment REQUIRES that TCP connections be 275 possible from the client system to the IMAP and Message Submission 276 ports on the servers, as well as from the Message Submission server 277 to the IMAP server. This may require configuring firewalls to 278 permit such usage. 280 Firewalls deployed in the network path MUST NOT damage protocol 281 traffic. In particular, both message submission and IMAP 282 connections from the client MUST be permitted. Firewalls MUST NOT 283 partially block extensions to these protocols, such as by allowing 284 one side of an extension negotiation, as doing so results in the two 286 Gellens [Page 6] Expires November 2007 287 sides being out of synch, with later failures. See [FIREWALLS] for 288 more discussion. 290 Application proxies, which are a not uncommon mechanism, are 291 discussed in [PROXIES]. 293 6.1 Firewall Traversal 295 An often-heard complaint from those attempting to deploy new 296 services within an organization is that the group responsible for 297 maintaining the firewall is unable or unwilling to open the required 298 ports. The group which owns the firewall, being charged with 299 organizational network security, is often reluctant to open firewall 300 ports without an understanding of the benefits and the security 301 implications of the new service. 303 The group wishing to deploy a new service is often tempted to bypass 304 the procedure and internal politics necessary to open the firewall 305 ports. A tempting kludge is to tunnel the new service over an 306 existing service that is already permitted to pass through the 307 firewall, typically HTTP on port 80 or sometimes SMTP on port 25. 308 Some of the downsides to this are discussed in [KLUDGE]. 310 Such bypass can appear to be immediately successful, since the new 311 service seems to deploy. However, assuming the network security 312 group is competent, when they become aware of the kludge, their 313 response is generally to block the violation of organizational 314 security policy. It is difficult to design an application-level 315 proxy/firewall which can provide such access control without 316 violating the transparency requirements of firewalls, as described 317 in [FIREWALLS]. Collateral damage is common in these circumstances. 318 The new service (which initially appeared to have been successfully 319 deployed) as well as those existing services which were leveraged to 320 tunnel the new service, becomes subject to arbitrary and 321 unpredictable failures. This encourages an adversarial relationship 322 between the two groups, which hinders attempts at resolution. 324 Even more serious is what happens if a vulnerability is discovered 325 in the new service. Until the vulnerability is corrected, the 326 network security group must disable both the new service and the 327 (typically mission-critical) existing service on which it is 328 layered. 330 An often-repeated truism is that any computer which is connected to 331 a network is insecure. Security and usefulness are both 332 considerations, with organizations making choices about achieving 333 acceptable measures in both areas. Deploying new services typically 334 requires deciding to permit access to the ports used by the service, 335 with appropriate protections. While the delay necessary to review 337 Gellens [Page 7] Expires November 2007 338 the implications of a new service may be frustrating, in the long 339 run it is likely to be less expensive than a kludge. 341 7 NATs 343 Many NAT boxes place lifetime limits on state, which has the effect 344 of aging out long-idle TCP connections. Since memory is relatively 345 cheap, there's little benefit in arbitrary timeouts. Instead, the 346 oldest unused connection can be recycled if memory or other 347 resources (such as IP addresses) become exhausted, allowing 348 connections to stay stay up forever when resources are available. 350 Any NAT boxes which are deployed between client and server systems 351 SHOULD be configured to have extremely long connection lifetimes. 352 Unlimited lifetimes are RECOMMENDED. 354 Note that IMAP and message submission clients will automatically 355 re-open TCP connections as needed, but it saves time, packets, and 356 processing to avoid the need to do so. Re-opening IMAP and message 357 submission connections generally incurs costs for authentication, 358 TLS negotiation, and server processing, as well as resetting of TCP 359 behavior such as windows. It is also ridiculously wasteful to force 360 clients to send NOOP commands just to maintain NAT state, especially 361 since this can defeat dormancy mode. 363 8 Security Considerations 365 There are numerous security considerations whenever an organization 366 chooses to make any of its services available via the Internet. 367 This includes email from mobile clients. 369 Sites concerned about email security should perform a threat 370 analysis, get relevant protections in place and then make a 371 conscious decision to open up this service. As discussed in section 372 6.1, piggybacking email traffic on the HTTP port in an attempt to 373 avoid making a firewall configuration change to explicitly permit 374 mobile email connections would bypass this important step and 375 reduces the overall security of the system. 377 Organizations might wish to purchase a messaging server which comes 378 with some indemnity and/or a messaging server which is used "on the 379 edge" by the organization that sells the server. 381 This document does not attempt to catalogue either the various risks 382 an organization might face or the numerous techniques which can be 383 used to protect against the risks. However, to help illustrate the 384 deployment considerations, a very small sample of some of the risks 386 Gellens [Page 8] Expires November 2007 387 and countermeasures appear below. 389 Some organizations are concerned that permitting direct access to 390 their mail servers via the Internet increases their vulnerability, 391 since a successful exploit against a mail server can potentially 392 expose all mail and authentication credentials stored on that 393 server, and can serve as an injection point for spam. In addition, 394 there are concerns over eavesdropping or modification of mail data 395 and authentication credentials. 397 A large number of approaches exist which can mitigate the risks 398 while allowing access to mail services via mobile clients. 400 Placing servers inside one or more DMZs (demilitarized zones, also 401 called perimeter networks) can protect the rest of the network from 402 a compromised server. An additional way to reduce the risk is to 403 store authentication credentials on a system which is not accessible 404 from the Internet, and which the servers within the DMZ can access 405 only by sending the credentials as received from the client and 406 receiving an authorized/not authorized response. Such isolation 407 reduces the ability of a compromised server to serve as a base for 408 attacking other network hosts. 410 Many additional techniques for further isolation exist, such as 411 having the DMZ IMAP server have no mail store of its own. When a 412 client connects to such a server, the DMZ IMAP server might contact 413 the authentication server and receive a ticket, which it passes to 414 the mail store in order to access the client's mail. In this way a 415 compromised IMAP server cannot be used to access the mail or 416 credentials for other users. 418 It is important to realize that simply throwing an extra box in 419 front of the mail servers, such as a gateway which may use HTTP or 420 any of a number of synchronization protocols to communicate with 421 clients, does not itself change the security aspects. By adding 422 such a gateway, the overall security of the system, and the 423 vulnerability of the mail servers, may remain unchanged or may be 424 significantly worsened. Isolation and indirection can be used to 425 protect against specific risks, but to be effective, such steps need 426 to be done after a threat analysis, and with understanding of the 427 issues involved. 429 Organizations SHOULD deploy servers which support the use of TLS for 430 all connections and which can be optionally configured to require 431 TLS. When TLS is used, it SHOULD be via the STARTTLS extensions 432 rather than the alternate port method. TLS can be an effective 433 measure to protect against specific threats, including eavesdropping 434 and alteration, of the traffic between the end-points. However, 435 just because TLS is deployed does not mean the system is "secure." 437 Gellens [Page 9] Expires November 2007 438 Attempts at bypassing current firewall policy when deploying new 439 services have serious risks, as discussed in section 6.1. 441 It's rare for a new service to not have associated security 442 considerations. Making email available to an organization's members 443 using mobile devices can offer significant benefits. 445 9 IANA Considerations 447 None. 449 10 Acknowledgments 451 Chris Newman and Phil Karn suggested very helpful text. Brian Ross 452 and Dave Cridland reviewed drafts and provided excellent 453 suggestions. 455 11 Normative References 457 [HOST-REQUIREMENTS] "Requirements for Internet Hosts -- 458 Communication Layers", R. Braden, RFC 1122, October 1989. 460 [KEYWORDS] "Key words for use in RFCs to Indicate Requirement 461 Levels", S. Bradner, RFC 2119, BCP 14, March 1997. 463 [IANA] IANA Port Number Registry, 464 466 [TCP] "Transmission Control Protocol", J. Postel, RFC 793, STD 7, 467 September 1981. 469 12 Informative References 471 [ACAP] "ACAP -- Application Configuration Access Protocol", C. 472 Newman, J.G. Myers, RFC 2244, November 1997. 474 [FIREWALLS] "Behavior of and Requirements for Internet Firewalls", 475 N. Freed, RFC 2979, October 2000. 477 [IMAP] "Internet Message Access Protocol -- Version 4rev1", M. 478 Crispin, RFC 3501, March 2003. 480 Gellens [Page 10] Expires November 2007 482 [KLUDGE] "On the use of HTTP as a Substrate", K. Moore, BCP 56, 483 February 2002. 485 [PROFILE] "Lemonade Profile", S. Maes, A. Melnikov, RFC 4550, June 486 2006. 488 [PROXIES] "Classical versus Transparent IP Proxies", M. Chatel, RFC 489 1919, March 1996. 491 [SUBMISSION] "Message Submission for Mail", R. Gellens, J. Klensin, 492 RFC 4409, April 2006. 494 13 Author's Address 496 Randall Gellens 497 QUALCOMM Incorporated 498 5775 Morehouse Drive 499 San Diego, CA 92121 500 randy@qualcomm.com 502 Appendix A: Changes from Previous Version 504 THIS SECTION TO BE REMOVED PRIOR TO PUBLICATION. 506 Changes made from version -07 to -08 as a result of IESG DISCUSS by 507 Cullen Jennings: 508 o Added text to section 4.1 explaining that the two-hour maximum 509 idle time is the default, while still permitting specific 510 configurations to exceed this to maximize dormancy. 511 o Changed ICMP soft errors from SHOULD NOT to MUST NOT. 513 Changes made from version -06 to -07 as a result of IESG DISCUSS by 514 Cullen Jennings: 515 o Removed prohibition on NATs closing connections just because of 516 how long they have been open, without taking any other factor 517 into account. Replaced with discussion on the issue and 518 recommendation that lemonade clients not allow connections to 519 remain idle for two hours, unless they have specific 520 configuration instructions to do so (e.g., in networks where 521 NATs and firewalls allow this, and the IMAP server permits it) 522 o Changed explicit note to permit all systems to terminate TCP 523 connections at any time based on local decisions 524 o Removed word "insurance" 526 Gellens [Page 11] Expires November 2007 527 Changes made from version -05 to -06 as a result of IESG DISCUSSes 528 by Jari Arkko, Lars Eggert, Cullen Jennings, and Magnus Westerlund: 529 o Clarified that "support" for TCP means availability of TCP to 530 applications, as opposed to deployment or use of TCP within a 531 network 532 o Clarified that an HTTP-only environment can offer webmail, which 533 may be email, but isn't lemonade email 534 o Added explicit note that end systems remain able to terminate 535 TCP connections at any time based on local decisions 536 o Made [FIREWALLS] an informative, not normative reference by 537 restating requirement 538 o Additional clarifications to make draft easier to read from a 539 non-Applications Area viewpoint 541 Changes made from version -04 to -05 as a result of IETF Last Call: 542 o Fixed some typos. 543 o Made first use of TCP into a reference. 545 Changes made from version -03 to -04 as a result of WG Last Call: 546 o New boilerplate text 547 o Wording tweaks from lemonade list (e.g., expanding contractions) 548 o Explcitly state that support for TCP is REQUIRED 549 o Correct reference in timeout text from PROFILE to IMAP 550 o Add RFC number to KEYWORDS reference (nit checker doesn't like 551 BCP number only) 552 o Move HOST-REQUIREMENTS reference to normative from informative 553 o Add TCP reference (since TCP support is REQUIRED) 554 o Add IMAP reference (for port number) 555 o Update PROFILE reference to RFC (from RFC Ed pub queue) 557 Intellectual Property Statement 559 The IETF takes no position regarding the validity or scope of any 560 Intellectual Property Rights or other rights that might be claimed 561 to pertain to the implementation or use of the technology described 562 in this document or the extent to which any license under such 563 rights might or might not be available; nor does it represent that 564 it has made any independent effort to identify any such rights. 565 Information on the procedures with respect to rights in RFC 566 documents can be found in BCP 78 and BCP 79. 568 Copies of IPR disclosures made to the IETF Secretariat and any 569 assurances of licenses to be made available, or the result of an 570 attempt made to obtain a general license or permission for the use 571 of such proprietary rights by implementers or users of this 572 specification can be obtained from the IETF on-line IPR repository 573 at http://www.ietf.org/ipr. 575 Gellens [Page 12] Expires November 2007 576 The IETF invites any interested party to bring to its attention any 577 copyrights, patents or patent applications, or other proprietary 578 rights that may cover technology that may be required to implement 579 this standard. Please address the information to the IETF at 580 ietf-ipr@ietf.org. 582 Full Copyright Statement 584 Copyright (C) The IETF Trust (2007). 586 This document is subject to the rights, licenses and restrictions 587 contained in BCP 78, and except as set forth therein, the authors 588 retain all their rights. 590 This document and the information contained herein are provided on 591 an "AS IS" basis and THE CONTRIBUTOR, THE ORGANIZATION HE/SHE 592 REPRESENTS OR IS SPONSORED BY (IF ANY), THE INTERNET SOCIETY, THE 593 IETF TRUST AND THE INTERNET ENGINEERING TASK FORCE DISCLAIM ALL 594 WARRANTIES, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO ANY 595 WARRANTY THAT THE USE OF THE INFORMATION HEREIN WILL NOT INFRINGE 596 ANY RIGHTS OR ANY IMPLIED WARRANTIES OF MERCHANTABILITY OR FITNESS 597 FOR A PARTICULAR PURPOSE. 599 Gellens [Page 13] Expires November 2007