7Network Working Group D. Crocker
8Request for Comments: 5598 Brandenburg InternetWorking
9Category: Informational July 2009
12 Internet Mail Architecture
16 Over its thirty-five-year history, Internet Mail has changed
17 significantly in scale and complexity, as it has become a global
18 infrastructure service. These changes have been evolutionary, rather
19 than revolutionary, reflecting a strong desire to preserve both its
20 installed base and its usefulness. To collaborate productively on
21 this large and complex system, all participants need to work from a
22 common view of it and use a common language to describe its
23 components and the interactions among them. But the many differences
24 in perspective currently make it difficult to know exactly what
25 another participant means. To serve as the necessary common frame of
26 reference, this document describes the enhanced Internet Mail
27 architecture, reflecting the current service.
31 This memo provides information for the Internet community. It does
32 not specify an Internet standard of any kind. Distribution of this
37 Copyright (c) 2009 IETF Trust and the persons identified as the
38 document authors. All rights reserved.
40 This document is subject to BCP 78 and the IETF Trust's Legal
41 Provisions Relating to IETF Documents in effect on the date of
42 publication of this document (http://trustee.ietf.org/license-info).
43 Please review these documents carefully, as they describe your rights
44 and restrictions with respect to this document.
46 This document may contain material from IETF Documents or IETF
47 Contributions published or made publicly available before November
48 10, 2008. The person(s) controlling the copyright in some of this
49 material may not have granted the IETF Trust the right to allow
50 modifications of such material outside the IETF Standards Process.
51 Without obtaining an adequate license from the person(s) controlling
52 the copyright in such materials, this document may not be modified
53 outside the IETF Standards Process, and derivative works of it may
58Crocker Informational [Page 1]
60RFC 5598 Email Architecture July 2009
63 not be created outside the IETF Standards Process, except to format
64 it for publication as an RFC or to translate it into languages other
69 1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . 3
70 1.1. History . . . . . . . . . . . . . . . . . . . . . . . . . 4
71 1.2. The Role of This Architecture . . . . . . . . . . . . . . 6
72 1.3. Document Conventions . . . . . . . . . . . . . . . . . . . 7
73 2. Responsible Actor Roles . . . . . . . . . . . . . . . . . . . 7
74 2.1. User Actors . . . . . . . . . . . . . . . . . . . . . . . 8
75 2.2. Message Handling Service (MHS) Actors . . . . . . . . . . 11
76 2.3. Administrative Actors . . . . . . . . . . . . . . . . . . 14
77 3. Identities . . . . . . . . . . . . . . . . . . . . . . . . . . 17
78 3.1. Mailbox . . . . . . . . . . . . . . . . . . . . . . . . . 17
79 3.2. Scope of Email Address Use . . . . . . . . . . . . . . . . 18
80 3.3. Domain Names . . . . . . . . . . . . . . . . . . . . . . . 19
81 3.4. Message Identifier . . . . . . . . . . . . . . . . . . . . 19
82 4. Services and Standards . . . . . . . . . . . . . . . . . . . . 21
83 4.1. Message Data . . . . . . . . . . . . . . . . . . . . . . . 24
84 4.1.4. Identity References in a Message . . . . . . . . . . . 25
85 4.2. User-Level Services . . . . . . . . . . . . . . . . . . . 29
86 4.3. MHS-Level Services . . . . . . . . . . . . . . . . . . . . 31
87 4.4. Transition Modes . . . . . . . . . . . . . . . . . . . . . 34
88 4.5. Implementation and Operation . . . . . . . . . . . . . . . 35
89 5. Mediators . . . . . . . . . . . . . . . . . . . . . . . . . . 35
90 5.1. Alias . . . . . . . . . . . . . . . . . . . . . . . . . . 37
91 5.2. ReSender . . . . . . . . . . . . . . . . . . . . . . . . . 38
92 5.3. Mailing Lists . . . . . . . . . . . . . . . . . . . . . . 39
93 5.4. Gateways . . . . . . . . . . . . . . . . . . . . . . . . . 41
94 5.5. Boundary Filter . . . . . . . . . . . . . . . . . . . . . 42
95 6. Considerations . . . . . . . . . . . . . . . . . . . . . . . . 42
96 6.1. Security Considerations . . . . . . . . . . . . . . . . . 42
97 6.2. Internationalization . . . . . . . . . . . . . . . . . . . 43
98 7. References . . . . . . . . . . . . . . . . . . . . . . . . . . 45
99 7.1. Normative References . . . . . . . . . . . . . . . . . . . 45
100 7.2. Informative References . . . . . . . . . . . . . . . . . . 47
101 Appendix A. Acknowledgments . . . . . . . . . . . . . . . . . . . 50
102 Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
114Crocker Informational [Page 2]
116RFC 5598 Email Architecture July 2009
121 Over its thirty-five-year history, Internet Mail has changed
122 significantly in scale and complexity, as it has become a global
123 infrastructure service. These changes have been evolutionary, rather
124 than revolutionary, reflecting a strong desire to preserve both its
125 installed base and its usefulness. Today, Internet Mail is
126 distinguished by many independent operators, many different
127 components for providing service to Users, as well as many different
128 components that transfer messages.
130 The underlying technical standards for Internet Mail comprise a rich
131 array of functional capabilities. These specifications form the
134 * Simple Mail Transfer Protocol (SMTP) ([RFC0821], [RFC2821],
135 [RFC5321]) moves a message through the Internet.
137 * Internet Mail Format (IMF) ([RFC0733], [RFC0822], [RFC2822],
138 [RFC5322]) defines a message object.
140 * Multipurpose Internet Mail Extensions (MIME) [RFC2045] defines
141 an enhancement to the message object that permits using
142 multimedia attachments.
144 Public collaboration on technical, operations, and policy activities
145 of email, including those that respond to the challenges of email
146 abuse, has brought a much wider range of participants into the
147 technical community. To collaborate productively on this large and
148 complex system, all participants need to work from a common view of
149 it and use a common language to describe its components and the
150 interactions among them. But the many differences in perspective
151 currently make it difficult to know exactly what another participant
154 It is the need to resolve these differences that motivates this
155 document, which describes the realities of the current system.
156 Internet Mail is the subject of ongoing technical, operations, and
157 policy work, and the discussions often are hindered by different
158 models of email-service design and different meanings for the same
161 To serve as the necessary common frame of reference, this document
162 describes the enhanced Internet Mail architecture, reflecting the
163 current service. The document focuses on:
170Crocker Informational [Page 3]
172RFC 5598 Email Architecture July 2009
175 * Capturing refinements to the email model
177 * Clarifying functional roles for the architectural components
179 * Clarifying identity-related issues, across the email service
181 * Defining terminology for architectural components and their
186 The first standardized architecture for networked email specified a
187 simple split between the user world, in the form of Message User
188 Agents (MUAs), and the transfer world, in the form of the Message
189 Handling Service (MHS), which is composed of Message Transfer Agents
190 (MTAs) [RFC1506]. The MHS accepts a message from one User and
191 delivers it to one or more other Users, creating a virtual MUA-to-MUA
192 exchange environment.
194 As shown in Figure 1, this architecture defines two logical layers of
195 interoperability. One is directly between Users. The other is among
196 the components along the transfer path. In addition, there is
197 interoperability between the layers, first when a message is posted
198 from the User to the MHS and later when it is delivered from the MHS
201 The operational service has evolved, although core aspects of the
202 service, such as mailbox addressing and message format style, remain
203 remarkably constant. The original distinction between the user level
204 and transfer level remains, but with elaborations in each. The term
205 "Internet Mail" is used to refer to the entire collection of user and
206 transfer components and services.
208 For Internet Mail, the term "end-to-end" usually refers to a single
209 posting and the set of deliveries that result from a single transit
210 of the MHS. A common exception is group dialogue that is mediated
211 through a Mailing List; in this case, two postings occur before
212 intended Recipients receive an Author's message, as discussed in
213 Section 2.1.4. In fact, some uses of email consider the entire email
214 service, including Author and Recipient, as a subordinate component.
215 For these services, "end-to-end" refers to points outside the email
216 service. Examples are voicemail over email [RFC3801], EDI
217 (Electronic Data Interchange) over email [RFC1767], and facsimile
218 over email [RFC4142].
226Crocker Informational [Page 4]
228RFC 5598 Email Architecture July 2009
232 ++================>| User |
235 +--------+ || +--------+ .
236 | User +==++=========>| User | .
237 +---+----+ || +--------+ .
245 +---+-----------------+------+------+---+
247 | .................>. . . |
249 | ........................>. . |
251 | ...............................>. |
253 | Message Handling Service (MHS) |
254 +---------------------------------------+
256 Legend: === lines indicate primary (possibly indirect)
258 ... lines indicate supporting transfers or roles
260 Figure 1: Basic Internet Mail Service Model
262 End-to-end Internet Mail exchange is accomplished by using a
263 standardized infrastructure with these components and
270 * An asynchronous sequence of point-to-point transfer mechanisms
272 * No requirement for prior arrangement between MTAs or between
273 Authors and Recipients
275 * No requirement for prior arrangement between point-to-point
276 transfer services over the open Internet
282Crocker Informational [Page 5]
284RFC 5598 Email Architecture July 2009
287 * No requirement for Author, Originator, or Recipients to be
288 online at the same time
290 The end-to-end portion of the service is the email object, called a
291 "message". Broadly, the message itself distinguishes control
292 information, for handling, from the Author's content.
294 A precept to the design of mail over the open Internet is permitting
295 User-to-User and MTA-to-MTA interoperability without prior, direct
296 arrangement between the independent administrative authorities
297 responsible for handling a message. All participants rely on having
298 the core services universally supported and accessible, either
299 directly or through Gateways that act as translators between Internet
300 Mail and email environments conforming to other standards. Given the
301 importance of spontaneity and serendipity in interpersonal
302 communications, not requiring such prearrangement between
303 participants is a core benefit of Internet Mail and remains a core
306 Within localized networks at the edge of the public Internet, prior
307 administrative arrangement often is required and can include access
308 control, routing constraints, and configuration of the information
309 query service. Although Recipient authentication has usually been
310 required for message access since the beginning of Internet Mail, in
311 recent years it also has been required for message submission. In
312 these cases, a server validates the client's identity, whether by
313 explicit security protocols or by implicit infrastructure queries to
314 identify "local" participants.
3161.2. The Role of This Architecture
318 An Internet service is an integration of related capabilities among
319 two or more participating nodes. The capabilities are accomplished
320 across the Internet by one or more protocols. What connects a
321 protocol to a service is an architecture. An architecture specifies
322 how the protocols implement the service by defining the logical
323 components of a service and the relationships among them. From that
324 logical view, a service defines what is being done, an architecture
325 defines where the pieces are (in relation to each other), and a
326 protocol defines how particular capabilities are performed.
328 As such, an architecture will more formally describe a service at a
329 relatively high level. A protocol that implements some portion of a
330 service will conform to the architecture to a greater or lesser
331 extent, depending on the pragmatic tradeoffs they make when trying to
332 implement the architecture in the context of real-world constraints.
333 Failure to precisely follow an architecture is not a failure of the
334 protocol, nor is failure to precisely cast a protocol a failure of
338Crocker Informational [Page 6]
340RFC 5598 Email Architecture July 2009
343 the architecture. Where a protocol varies from the architecture, it
344 will of course be appropriate for it to explain the reason for the
345 variance. However, such variance is not a mark against a protocol:
346 Happily, the IETF prefers running code to architectural purity.
348 In this particular case, this architecture attempts to define the
349 logical components of Internet email and does so post hoc, trying to
350 capture the architectural principles that the current email protocols
351 embody. To different extents, email protocols will conform to this
352 architecture more or less well. Insofar as this architecture differs
353 from those protocols, the reasons are generally well understood and
354 are required for interoperation. The differences are not a sign that
355 protocols need to be fixed. However, this architecture is a best
356 attempt at a logical model of Internet email, and insofar as new
357 protocol development varies from this architecture, it is necessary
358 for designers to understand those differences and explain them
3611.3. Document Conventions
363 References to structured fields of a message use a two-part dotted
364 notation. The first part cites the document that contains the
365 specification for the field, and the second part is the name of the
366 field. Hence <RFC5322.From> is the IMF From: header field in an
367 email content header, and <RFC5321.MailFrom> is the address in the
368 SMTP "Mail From" command.
370 When occurring without the IMF (RFC 5322) qualifier, header field
371 names are shown with a colon suffix. For example, From:.
373 References to labels for actors, functions or components have the
374 first letter capitalized.
3762. Responsible Actor Roles
378 Internet Mail is a highly distributed service, with a variety of
379 Actors playing different roles. These Actors fall into three basic
384 * Message Handling Service (MHS)
386 * ADministrative Management Domain (ADMD)
394Crocker Informational [Page 7]
396RFC 5598 Email Architecture July 2009
399 Although related to a technical architecture, the focus on Actors
400 concerns participant responsibilities, rather than functionality of
401 modules. For that reason, the labels used are different from those
402 used in classic diagrams of email architecture.
406 Users are the sources and sinks of messages. Users can be people,
407 organizations, or processes. They can have an exchange that
408 iterates, and they can expand or contract the set of Users that
409 participate in a set of exchanges. In Internet Mail, there are four
420 Figure 2 shows the primary and secondary flows of messages among
421 them. As a pragmatic heuristic: User Actors can generate, modify, or
422 look at the whole message.
450Crocker Informational [Page 8]
452RFC 5598 Email Architecture July 2009
456 || Author ||<..................................<..
458 || || || ++===========++ .
459 || || ++====>|| Recipient || .
460 || || ++=====+=====++ .
462 || || ..........................>.+
464 || || ................... .
467 || || +-----------+ ++=====+=====++ .
468 || ++========>| Mediator +===>|| Recipient || .
469 || +-----+-----+ ++=====+=====++ .
471 || ..................+.......>.+
473 || ..............+.................. .
476 +-----------+ +-----------+ ++=====+=====++ .
477 | Mediator +===>| Mediator +===>|| Recipient || .
478 +-----+-----+ +-----+-----+ ++=====+=====++ .
480 .................+.................+.......>..
482 Legend: === lines indicate primary (possibly indirect)
484 ... lines indicate supporting transfers or roles
486 Figure 2: Relationships among User Actors
488 From a User's perspective, all message-transfer activities are
489 performed by a monolithic Message Handling Service (MHS), even though
490 the actual service can be provided by many independent organizations.
491 Users are customers of this unified service.
493 Whenever any MHS Actor sends information back to an Author or
494 Originator in the sequence of handling a message, that Actor is a
499 The Author is responsible for creating the message, its contents, and
500 its list of Recipient addresses. The MHS transfers the message from
501 the Author and delivers it to the Recipients. The MHS has an
502 Originator role (Section 2.2.1) that correlates with the Author role.
506Crocker Informational [Page 9]
508RFC 5598 Email Architecture July 2009
513 The Recipient is a consumer of the delivered message. The MHS has a
514 Receiver role (Section 2.2.4) that correlates with the Recipient
515 role. This is labeled Recv in Figure 3.
517 Any Recipient can close the user-communication loop by creating and
518 submitting a new message that replies to the Author. An example of
519 an automated form of reply is the Message Disposition Notification
520 (MDN), which informs the Author about the Recipient's handling of the
521 message. (See Section 4.1.)
525 Also called "Bounce Handler", the Return Handler is a special form of
526 Recipient tasked with servicing notifications generated by the MHS as
527 it transfers or delivers the message. (See Figure 3.) These notices
528 can be about failures or completions and are sent to an address that
529 is specified by the Originator. This Return Handling address (also
530 known as a Return Address) might have no visible characteristics in
531 common with the address of the Author or Originator.
535 A Mediator receives, aggregates, reformulates, and redistributes
536 messages among Authors and Recipients who are the principals in
537 (potentially) protracted exchanges. This activity is easily confused
538 with the underlying MHS transfer exchanges. However, each serves
539 very different purposes and operates in very different ways.
541 When mail is delivered to the Mediator specified in the
542 RFC5321.RcptTo command for the original message, the MHS handles it
543 the same way as for any other Recipient. In particular, the MHS sees
544 each posting and delivery activity between sources and sinks as
545 independent; it does not see subsequent re-posting as a continuation
546 of a process. Because the Mediator originates messages, it can
547 receive replies. Hence, when submitting a reformulated message, the
548 Mediator is an Author, albeit an Author actually serving as an agent
549 of one or more other Authors. So a Mediator really is a full-fledged
550 User. Mediators are considered extensively in Section 5.
552 A Mediator attempts to preserve the original Author's information in
553 the message it reformulates but is permitted to make meaningful
554 changes to the message content or envelope. The MHS sees a new
555 message, but Users receive a message that they interpret as being
556 from, or at least initiated by, the Author of the original message.
557 The role of a Mediator is not limited to merely connecting other
558 participants; the Mediator is responsible for the new message.
562Crocker Informational [Page 10]
564RFC 5598 Email Architecture July 2009
567 A Mediator's role is complex and contingent, for example, modifying
568 and adding content or regulating which Users are allowed to
569 participate and when. The common example of this role is a group
570 Mailing List. In a more complex use, a sequence of Mediators could
571 perform a sequence of formal steps, such as reviewing, modifying, and
572 approving a purchase request.
574 A Gateway is a particularly interesting form of Mediator. It is a
575 hybrid of User and Relay that connects heterogeneous mail services.
576 Its purpose is to emulate a Relay. For a detailed discussion, see
5792.2. Message Handling Service (MHS) Actors
581 The Message Handling Service (MHS) performs a single end-to-end
582 transfer on behalf of the Author to reach the Recipient addresses
583 specified in the original RFC5321.RcptTo commands. Exchanges that
584 are either mediated or iterative and protracted, such as those used
585 for collaboration over time, are handled by the User Actors, not by
586 the MHS Actors. As a pragmatic heuristic MHS Actors generate,
587 modify, or look at only transfer data, rather than the entire
590 Figure 3 shows the relationships among transfer participants in
591 Internet Mail. Although it shows the Originator (labeled Origin) as
592 distinct from the Author, and Receiver (labeled Recv) as distinct
593 from Recipient, each pair of roles usually has the same Actor.
594 Transfers typically entail one or more Relays. However, direct
595 delivery from the Originator to Receiver is possible. Intra-
596 organization mail services usually have only one Relay.
618Crocker Informational [Page 11]
620RFC 5598 Email Architecture July 2009
623 ++==========++ ++===========++
624 || Author || || Recipient ||
625 ++====++====++ +--------+ ++===========++
629 +---------+ . . +---++---+
631 /--+---------+----------------------------+--------+----\
633 | | Origin +<...... .................+ Recv | |
635 | +---++----+ . +--------+ |
637 | || ..............+.................. || |
639 | +-------+-+ +--+------+ +-+--++---+ |
640 | | Relay +=======>| Relay +=======>| Relay | |
641 | +---------+ +----++---+ +---------+ |
646 | | Gateway +-->... |
648 \-------------------------------------------------------/
650 Legend: === and || lines indicate primary (possibly
651 indirect) transfers or roles
652 ... lines indicate supporting transfers or roles
654 Figure 3: Relationships among MHS Actors
658 The Originator ensures that a message is valid for posting and then
659 submits it to a Relay. A message is valid if it conforms to both
660 Internet Mail standards and local operational policies. The
661 Originator can simply review the message for conformance and reject
662 it if it finds errors, or it can create some or all of the necessary
663 information. In effect, the Originator is responsible for the
664 functions of the Mail Submission Agent.
666 The Originator operates with dual allegiance. It serves the Author
667 and can be the same entity. But its role in assuring validity means
668 that it also represents the local operator of the MHS, that is, the
669 local ADministrative Management Domain (ADMD).
674Crocker Informational [Page 12]
676RFC 5598 Email Architecture July 2009
679 The Originator also performs any post-submission, Author-related
680 administrative tasks associated with message transfer and delivery.
681 Notably, these tasks pertain to sending error and delivery notices,
682 enforcing local policies, and dealing with messages from the Author
683 that prove to be problematic for the Internet. The Originator is
684 accountable for the message content, even when it is not responsible
685 for it. The Author creates the message, but the Originator handles
686 any transmission issues with it.
690 The Relay performs MHS-level transfer-service routing and store-and-
691 forward by transmitting or retransmitting the message to its
692 Recipients. The Relay adds trace information [RFC2505] but does not
693 modify the envelope information or the message content semantics. It
694 can modify message content representation, such as changing the form
695 of transfer encoding from binary to text, but only as required to
696 meet the capabilities of the next hop in the MHS.
698 A Message Handling System (MHS) network consists of a set of Relays.
699 This network is above any underlying packet-switching network that
700 might be used and below any Gateways or other Mediators.
702 In other words, email scenarios can involve three distinct
703 architectural layers, each providing its own type of data of store-
712 The bottom layer is the Internet's IP service. The most basic email
713 scenarios involve Relays and Switches.
715 When a Relay stops attempting to transfer a message, it becomes an
716 Author because it sends an error message to the Return Address. The
717 potential for looping is avoided by omitting a Return Address from
722 A Gateway is a hybrid of User and Relay that connects heterogeneous
723 mail services. Its purpose is to emulate a Relay and the closer it
724 comes to this, the better. A Gateway operates as a User when it
725 needs the ability to modify message content.
730Crocker Informational [Page 13]
732RFC 5598 Email Architecture July 2009
735 Differences between mail services can be as small as minor syntax
736 variations, but they usually encompass significant, semantic
737 distinctions. One difference could be email addresses that are
738 hierarchical and machine-specific rather than a flat, global
739 namespace. Another difference could be support for text-only content
740 or multimedia. Hence the Relay function in a Gateway presents a
741 significant design challenge if the resulting performance is to be
742 seen as nearly seamless. The challenge is to ensure User-to-User
743 functionality between the services, despite differences in their
744 syntax and semantics.
746 The basic test of Gateway design is whether an Author on one side of
747 a Gateway can send a useful message to a Recipient on the other side,
748 without requiring changes to any components in the Author's or
749 Recipient's mail services other than adding the Gateway. To each of
750 these otherwise independent services, the Gateway appears to be a
751 native participant. But the ultimate test of Gateway design is
752 whether the Author and Recipient can sustain a dialogue. In
753 particular, can a Recipient's MUA automatically formulate a valid
754 Reply that will reach the Author?
758 The Receiver performs final delivery or sends the message to an
759 alternate address. It can also perform filtering and other policy
760 enforcement immediately before or after delivery.
7622.3. Administrative Actors
764 Administrative Actors can be associated with different organizations,
765 each with its own administrative authority. This operational
766 independence, coupled with the need for interaction between groups,
767 provides the motivation to distinguish among ADministrative
768 Management Domains (ADMDs). Each ADMD can have vastly different
769 operating policies and trust-based decision-making. One obvious
770 example is the distinction between mail that is exchanged within an
771 organization and mail that is exchanged between independent
772 organizations. The rules for handling both types of traffic tend to
773 be quite different. That difference requires defining the boundaries
774 of each, and this requires the ADMD construct.
776 Operation of Internet Mail services is carried out by different
777 providers (or operators). Each can be an independent ADMD. This
778 independence of administrative decision-making defines boundaries
779 that distinguish different portions of the Internet Mail service. A
780 department that operates a local Relay, an IT department that
781 operates an enterprise Relay, and an ISP that operates a public
782 shared email service can be configured into many combinations of
786Crocker Informational [Page 14]
788RFC 5598 Email Architecture July 2009
791 administrative and operational relationships. Each is a distinct
792 ADMD, potentially having a complex arrangement of functional
793 components. Figure 4 depicts relationships among ADMDs. The benefit
794 of the ADMD construct is that it facilitates discussion about
795 designs, policies, and operations that need to distinguish between
796 internal issues and external ones.
798 The architectural impact of the need for boundaries between ADMDs is
799 discussed in [Tussle]. Most significant is that the entities
800 communicating across ADMD boundaries typically have the added burden
801 of enforcing organizational policies concerning external
802 communications. At a more mundane level, routing mail between ADMDs
803 can be an issue, such as needing to route mail between organizational
804 partners over specially trusted paths.
806 These are three basic types of ADMDs:
808 Edge: Independent transfer services in networks at the edge of
809 the open Internet Mail service.
811 Consumer: Might be a type of Edge service, as is common for web-
814 Transit: Mail Service Providers (MSPs) that offer value-added
815 capabilities for Edge ADMDs, such as aggregation and
818 The mail-level transit service is different from packet-level
819 switching. End-to-end packet transfers usually go through
820 intermediate routers; email exchange across the open Internet can be
821 directly between the Boundary MTAs of Edge ADMDs. This distinction
822 between direct and indirect interaction highlights the differences
823 discussed in Section 2.2.2.
842Crocker Informational [Page 15]
844RFC 5598 Email Architecture July 2009
847 +--------+ +---------+ +-------+ +-----------+
848 | ADMD1 |<===>| ADMD2 |<===>| ADMD3 |<===>| ADMD4 |
849 | ----- | | ----- | | ----- | | ----- |
851 | Author | | | | | | Recipient |
854 | Edge..+....>|.Transit.+....>|-Edge..+....>|..Consumer |
856 +--------+ +---------+ +-------+ +-----------+
858 Legend: === lines indicate primary (possibly indirect)
860 ... lines indicate supporting transfers or roles
862 Figure 4: Administrative Domain (ADMD) Example
864 Edge networks can use proprietary email standards internally.
865 However, the distinction between Transit network and Edge network
866 transfer services is significant because it highlights the need for
867 concern over interaction and protection between independent
868 administrations. In particular, this distinction calls for
869 additional care in assessing the transitions of responsibility and
870 the accountability and authorization relationships among participants
873 The interactions of ADMD components are subject to the policies of
874 that domain, which cover concerns such as these:
882 * Content evaluation and modification
884 These policies can be implemented in different functional components,
885 according to the needs of the ADMD. For example, see [RFC5068].
887 Consumer, Edge, and Transit services can be offered by providers that
888 operate component services or sets of services. Further, it is
889 possible for one ADMD to host services for other ADMDs.
898Crocker Informational [Page 16]
900RFC 5598 Email Architecture July 2009
903 These are common examples of ADMDs:
905 Enterprise Service Providers:
907 These ADMDs operate the internal data and/or the mail services
908 within an organization.
910 Internet Service Providers (ISP):
912 These ADMDs operate the underlying data communication services,
913 which are used by one or more Relay and User. ISPs are not
914 responsible for performing email functions, but they can provide
915 an environment in which those functions can be performed.
917 Mail Service Providers:
919 These ADMDs operate email services, such as for consumers or
922 Practical operational concerns demand that providers be involved in
923 administration and enforcement issues. This involvement can extend
924 to operators of lower-level packet services.
928 The forms of identity used by Internet Mail are: mailbox, domain
929 name, message-ID, and ENVID (envelope identifier). Each is globally
934 "A mailbox receives mail. It is a conceptual entity that does not
935 necessarily pertain to file storage." [RFC5322]
937 A mailbox is specified as an Internet Mail address <addr-spec>. It
938 has two distinct parts, separated by an at-sign (@). The right side
939 is a globally interpreted domain name associated with an ADMD.
940 Domain names are discussed in Section 3.3. Formal Internet Mail
941 addressing syntax can support source routes to indicate the path
942 through which a message ought to be sent. The use of source routes
943 is not common and has been deprecated in [RFC5321].
945 The portion to the left of the at-sign contains a string that is
946 globally opaque and is called the <local-part>. It is interpreted
947 only by the entity specified by the address's domain name. Except as
948 noted later in this section, all other entities treat the
949 <local-part> as an uninterpreted literal string and preserve all
954Crocker Informational [Page 17]
956RFC 5598 Email Architecture July 2009
959 of its original details. As such, its public distribution is
960 equivalent to sending a Web browser "cookie" that is only interpreted
961 upon being returned to its creator.
963 Some local-part values have been standardized for contacting
964 personnel at an organization. These names cover common operations
965 and business functions [RFC2142].
967 It is common for sites to have local structuring conventions for the
968 left-hand side, <local-part>, of an <addr-spec>. This permits sub-
969 addressing, such as for distinguishing different discussion groups
970 used by the same participant. However, it is worth stressing that
971 these conventions are strictly private to the User's organization and
972 are not interpreted by any domain except the one listed in the right
973 side of the <addr-spec>. The exceptions are those specialized
974 services that conform to public, standardized conventions, as noted
977 Basic email addressing defines the <local-part> as being globally
978 opaque. However, there are some uses of email that add a
979 standardized, global schema to the value, such as between an Author
980 and a Gateway. The <local-part> details remain invisible to the
981 public email transfer infrastructure, but provide addressing and
982 handling instructions for further processing by the Gateway.
983 Standardized examples of these conventions are the telephone
984 numbering formats for the Voice Profile for Internet Mail (VPIM)
987 +16137637582@vpim.example.com,
989 and iFax ([RFC3192], [RFC4143] such as:
991 FAX=+12027653000/T33S=1387@ifax.example.com.
9933.2. Scope of Email Address Use
995 Email addresses are being used far beyond their original role in
996 email transfer and delivery. In practical terms, an email address
997 string has become the common identifier for representing online
998 identity. Hence, it is essential to be clear about both the nature
999 and role of an identity string in a particular context and the entity
1000 responsible for setting that string. For example, see Sections
1001 4.1.4, 4.3.3, and 5.
1010Crocker Informational [Page 18]
1012RFC 5598 Email Architecture July 2009
1017 A domain name is a global reference to an Internet resource, such as
1018 a host, a service, or a network. A domain name usually maps to one
1019 or more IP Addresses. Conceptually, the name can encompass an
1020 organization, a collection of machines integrated into a homogeneous
1021 service, or a single machine. A domain name can be administered to
1022 refer to an individual User, but this is not common practice. The
1023 name is structured as a hierarchical sequence of labels, separated by
1024 dots (.), with the top of the hierarchy being on the right end of the
1025 sequence. There can be many names in the sequence -- that is, the
1026 depth of the hierarchy can be substantial. Domain names are defined
1027 and operated through the Domain Name System (DNS) ([RFC1034],
1028 [RFC1035], [RFC2181]).
1030 When not part of a mailbox address, a domain name is used in Internet
1031 Mail to refer to the ADMD or to the host that took action upon the
1032 message, such as providing the administrative scope for a message
1033 identifier or performing transfer processing.
10353.4. Message Identifier
1037 There are two standardized tags for identifying messages: Message-ID:
1038 and ENVID. A Message-ID: pertains to content, and an ENVID pertains
1043 IMF provides for, at most, a single Message-ID:. The Message-ID: for
1044 a single message, which is a user-level IMF tag, has a variety of
1045 uses including threading, aiding identification of duplicates, and
1046 DSN (Delivery Status Notification) tracking. The Originator assigns
1047 the Message-ID:. The Recipient's ADMD is the intended consumer of
1048 the Message-ID:, although any Actor along the transfer path can use
1051 Message-ID: is globally unique. Its format is similar to that of a
1052 mailbox, with two distinct parts separated by an at-sign (@).
1053 Typically, the right side specifies the ADMD or host that assigns the
1054 identifier, and the left side contains a string that is globally
1055 opaque and serves to uniquely identify the message within the domain
1056 referenced on the right side. The duration of uniqueness for the
1057 message identifier is undefined.
1059 When a message is revised in any way, the decision whether to assign
1060 a new Message-ID: requires a subjective assessment to determine
1061 whether the editorial content has been changed enough to constitute a
1062 new message. [RFC5322] states that "a message identifier pertains to
1066Crocker Informational [Page 19]
1068RFC 5598 Email Architecture July 2009
1071 exactly one version of a particular message; subsequent revisions to
1072 the message each receive new message identifiers." Yet experience
1073 suggests that some flexibility is needed. An impossible test is
1074 whether the Recipient will consider the new message to be equivalent
1075 to the old one. For most components of Internet Mail, there is no
1076 way to predict a specific Recipient's preferences on this matter.
1077 Both creating and failing to create a new Message-ID: have their
1080 Here are some guidelines and examples:
1082 o If a message is changed only in form, such as character encoding,
1083 it is still the same message.
1085 o If a message has minor additions to the content, such as a Mailing
1086 List tag at the beginning of the RFC5322.Subject header field, or
1087 some Mailing List administrative information added to the end of
1088 the primary body part text, it is probably the same message.
1090 o If a message has viruses deleted from it, it is probably the same
1093 o If a message has offensive words deleted from it, some Recipients
1094 will consider it the same message, but some will not.
1096 o If a message is translated into a different language, some
1097 Recipients will consider it the same message, but some will not.
1099 o If a message is included in a digest of messages, the digest
1100 constitutes a new message.
1102 o If a message is forwarded by a Recipient, what is forwarded is a
1105 o If a message is "redirected", such as using IMF "Resent-*" header
1106 fields, some Recipients will consider it the same message, but
1109 The absence of both objective, precise criteria for regenerating a
1110 Message-ID: and strong protection associated with the string means
1111 that the presence of an ID can permit an assessment that is
1112 marginally better than a heuristic, but the ID certainly has no value
1113 on its own for strict formal reference or comparison. For that
1114 reason, the Message-ID: is not intended to be used for any function
1115 that has security implications.
1122Crocker Informational [Page 20]
1124RFC 5598 Email Architecture July 2009
1129 The ENVID (envelope identifier) can be used for message-tracking
1130 purposes ([RFC3885], [RFC3464]) concerning a single posting/delivery
1131 transfer. The ENVID labels a single transit of the MHS by a specific
1132 message. So, the ENVID is used for one message posting until that
1133 message is delivered. A re-posting of the message, such as by a
1134 Mediator, does not reuse that ENVID, but can use a new one, even
1135 though the message might legitimately retain its original
1138 The format of an ENVID is free form. Although its creator might
1139 choose to impose structure on the string, none is imposed by Internet
1140 standards. By implication, the scope of the string is defined by the
1141 domain name of the Return Address.
11434. Services and Standards
1145 The Internet Mail architecture comprises six basic types of
1146 functionality, which are arranged to support a store-and-forward
1147 service. As shown in Figure 5, each type can have multiple
1148 instances, some of which represent specialized roles. This section
1149 considers the activities and relationships among these components,
1150 and the Internet Mail standards that apply to them.
1154 Message User Agent (MUA)
1158 Recipient MUA (rMUA)
1160 Message Submission Agent (MSA)
1162 Author-focused MSA functions (aMSA)
1164 MHS-focused MSA functions (hMSA)
1166 Message Transfer Agent (MTA)
1168 Message Delivery Agent (MDA)
1170 Recipient-focused MDA functions (rMDA)
1172 MHS-focused MDA functions (hMDA)
1178Crocker Informational [Page 21]
1180RFC 5598 Email Architecture July 2009
1189 This figure shows function modules and the standardized protocols
1234Crocker Informational [Page 22]
1236RFC 5598 Email Architecture July 2009
1241 ...........++ aMUA ||<............................+ Disp |
1244 . local,imap}| |{smtp,submission .
1245 . +-----+ | | +--------+ .
1246 . | aMS |<---+ | ........................>| Return | .
1247 . +-----+ | . +--------+ .
1248 . | . ***************** ^ .
1249 . +-----V-.----*------------+ * . .
1250 . MSA | +-------+ * +------+ | * . .
1251 . | | aMSA +-(S)->| hMSA | | * . .
1252 . | +-------+ * +--+---+ | * . .
1253 V +------------*------+-----+ * . .
1254 //==========\\ * V {smtp * . .
1255 || MESSAGE || * +------+ * //===+===\\ .
1256 ||----------|| MHS * | MTA | * || dsn || .
1257 || ENVELOPE || * +--+---+ * \\=======// .
1258 || smtp || * V {smtp * ^ ^ .
1259 || CONTENT || * +------+ * . . //==+==\\
1260 || imf || * | MTA +....*...... . || mdn ||
1261 || mime || * +--+---+ * . \\=====//
1262 \\==========// * smtp}| {local * . ^
1263 . MDA * | {lmtp * . .
1264 . +----------------+------V-----+ * . .
1265 . | +----------+ * +------+ | * . .
1266 . | | | * | | +..*.......... .
1267 . | | rMDA |<-(D)--+ hMDA | | * .
1268 . | | | * | | |<.*........ .
1269 . | +-+------+-+ * +------+ | * . .
1270 . +------+---------*------------+ * . .
1271 . smtp,local}| ***************** . .
1273 . +-----+ //===+===\\ .
1274 . | rMS | || sieve || .
1275 . +--+--+ \\=======// .
1276 . |{imap,pop,local ^ .
1278 . ++==========++ . .
1280 .......>|| rMUA ++........................... .
1281 || ++...................................
1284 Legend: --- lines indicate primary (possibly indirect)
1286 === boxes indicate data objects
1290Crocker Informational [Page 23]
1292RFC 5598 Email Architecture July 2009
1295 ... lines indicate supporting transfers or roles
1296 *** lines indicate aggregated service
1298 Figure 5: Protocols and Services
1302 The purpose of the Message Handling System (MHS) is to exchange an
1303 IMF message object among participants [RFC5322]. All of its
1304 underlying mechanisms serve to deliver that message from its Author
1305 to its Recipients. A message can be explicitly labeled as to its
1308 A message comprises a transit-handling envelope and the message
1309 content. The envelope contains information used by the MHS. The
1310 content is divided into a structured header and the body. The header
1311 comprises transit-handling trace information and structured fields
1312 that are part of the Author's message content. The body can be
1313 unstructured lines of text or a tree of multimedia subordinate
1314 objects, called "body-parts" or, popularly, "attachments".
1315 [RFC2045], [RFC2046], [RFC2047], [RFC4288], [RFC4289], [RFC2049].
1317 In addition, Internet Mail has a few conventions for special control
1320 Delivery Status Notification (DSN):
1322 A Delivery Status Notification (DSN) is a message that can be
1323 generated by the MHS (MSA, MTA, or MDA) and sent to the
1324 RFC5321.MailFrom address. MDA and MTA are shown as sources of
1325 DSNs in Figure 5, and the destination is shown as Returns. DSNs
1326 provide information about message transit, such as transfer errors
1327 or successful delivery [RFC3461].
1329 Message Disposition Notification (MDN):
1331 A Message Disposition Notification (MDN) is a message that
1332 provides information about post-delivery processing, such as
1333 indicating that the message has been displayed [RFC3798] or the
1334 form of content that can be supported [RFC3297]. It can be
1335 generated by an rMUA and is sent to the
1336 Disposition-Notification-To addresses. The mailbox for this is
1337 shown as Disp in Figure 5.
1346Crocker Informational [Page 24]
1348RFC 5598 Email Architecture July 2009
1351 Message Filtering (SIEVE):
1353 Sieve is a scripting language used to specify conditions for
1354 differential handling of mail, typically at the time of delivery
1355 [RFC5228]. Scripts can be conveyed in a variety of ways, such as
1356 a MIME part in a message. Figure 5 shows a Sieve script going
1357 from the rMUA to the MDA. However, filtering can be done at many
1358 different points along the transit path, and any one or more of
1359 them might be subject to Sieve directives, especially within a
1360 single ADMD. Figure 5 shows only one relationship, for (relative)
1365 Internet Mail has a fragmented framework for transit-related handling
1366 information. Information that is used directly by the MHS is called
1367 the "envelope". It directs handling activities by the transfer
1368 service and is carried in transfer-service commands. That is, the
1369 envelope exists in the transfer protocol SMTP [RFC5321].
1371 Trace information, such as RFC5322.Received, is recorded in the
1372 message header and is not subsequently altered [RFC5322].
1376 Header fields are attribute name/value pairs that cover an extensible
1377 range of email-service parameters, structured user content, and user
1378 transaction meta-information. The core set of header fields is
1379 defined in [RFC5322]. It is common practice to extend this set for
1380 different applications. Procedures for registering header fields are
1381 defined in [RFC3864]. An extensive set of existing header field
1382 registrations is provided in [RFC4021].
1384 One danger of placing additional information in header fields is that
1385 Gateways often alter or delete them.
1389 The body of a message might be lines of ASCII text or a
1390 hierarchically structured composition of multimedia body part
1391 attachments using MIME ([RFC2045], [RFC2046], [RFC2047], [RFC4288],
13944.1.4. Identity References in a Message
1396 Table 1 lists the core identifiers present in a message during
1402Crocker Informational [Page 25]
1404RFC 5598 Email Architecture July 2009
1407 +----------------------+----------------+---------------------------+
1408 | Layer | Field | Set By |
1409 +----------------------+----------------+---------------------------+
1410 | Message Body | MIME Header | Author |
1411 | Message header | From: | Author |
1413 | | Sender: | Originator |
1414 | | Reply-To: | Author |
1415 | | To:, CC:, BCC: | Author |
1416 | | Message-ID: | Originator |
1417 | | Received: | Originator, Relay, |
1419 | | Return-Path: | MDA, from MailFrom |
1420 | | Resent-*: | Mediator |
1421 | | List-Id: | Mediator |
1422 | | List-*: | Mediator |
1423 | SMTP | HELO/EHLO | Latest Relay Client |
1424 | | ENVID | Originator |
1425 | | MailFrom | Originator |
1426 | | RcptTo | Author |
1427 | | ORCPT | Originator |
1428 | IP | Source Address | Latest Relay Client |
1429 +----------------------+----------------+---------------------------+
1432 Layer - The part of the email architecture that uses the
1435 Field - The protocol construct that contains the identifier.
1437 Set By - The Actor role responsible for specifying the identifier
1438 value (and this can be different from the Actor that performs the
1439 fill-in function for the protocol construct).
1441 Table 1: Layered Identities
1443 These are the most common address-related fields:
1445 RFC5322.From: Set by - Author
1447 Names and addresses for Authors of the message content are listed
1458Crocker Informational [Page 26]
1460RFC 5598 Email Architecture July 2009
1463 RFC5322.Reply-To: Set by - Author
1465 If a Recipient sends a reply message that would otherwise use the
1466 RFC5322.From field addresses in the original message, the
1467 addresses in the RFC5322.Reply-To field are used instead. In
1468 other words, this field overrides the From: field for responses
1471 RFC5322.Sender: Set by - Originator
1473 This field specifies the address responsible for submitting the
1474 message to the transfer service. This field can be omitted if it
1475 contains the same address as RFC5322.From. However, omitting this
1476 field does not mean that no Sender is specified; it means that
1477 that header field is virtual and that the address in the From:
1478 field is to be used.
1480 Specification of the notifications Return Addresses, which are
1481 contained in RFC5321.MailFrom, is made by the RFC5322.Sender.
1482 Typically, the Return address is the same as the Sender address.
1483 However, some usage scenarios require it to be different.
1485 RFC5322.To/.CC: Set by - Author
1487 These fields specify MUA Recipient addresses. However, some or
1488 all of the addresses in these fields might not be present in the
1489 RFC5321.RcptTo commands.
1491 The distinction between To and CC is subjective. Generally, a To
1492 addressee is considered primary and is expected to take action on
1493 the message. A CC addressee typically receives a copy as a
1496 RFC5322.BCC: Set by - Author
1498 A copy of the message might be sent to an addressee whose
1499 participation is not to be disclosed to the RFC5322.To or
1500 RFC5322.CC Recipients and, usually, not to the other BCC
1501 Recipients. The BCC: header field indicates a message copy to
1502 such a Recipient. Use of this field is discussed in [RFC5322].
1504 RFC5321.HELO/.EHLO: Set by - Originator, MSA, MTA
1506 Any SMTP client -- including Originator, MSA, or MTA -- can
1507 specify its hosting domain identity for the SMTP HELO or EHLO
1514Crocker Informational [Page 27]
1516RFC 5598 Email Architecture July 2009
1519 RFC3461.ENVID: Set by - Originator
1521 The MSA can specify an opaque string, to be included in a DSN, as
1522 a means of assisting the Return Address Recipient in identifying
1523 the message that produced a DSN or message tracking.
1525 RFC5321.MailFrom: Set by - Originator
1527 This field is an end-to-end string that specifies an email address
1528 for receiving return control information, such as returned
1529 messages. The name of this field is misleading, because it is not
1530 required to specify either the Author or the Actor responsible for
1531 submitting the message. Rather, the Actor responsible for
1532 submission specifies the RFC5321.MailFrom address. Ultimately,
1533 the simple basis for deciding which address needs to be in the
1534 RFC5321.MailFrom field is to determine which address is to be
1535 informed about transfer-level problems (and possibly successes).
1537 RFC5321.RcptTo: Set by - Author, Final MTA, MDA
1539 This field specifies the MUA mailbox address of a Recipient. The
1540 string might not be visible in the message content header. For
1541 example, the IMF destination address header fields, such as
1542 RFC5322.To, might specify a Mailing List mailbox, while the
1543 RFC5321.RcptTo address specifies a member of that list.
1545 RFC5321.ORCPT: Set by - Originator.
1547 This is an optional parameter to the RCPT command, indicating the
1548 original address to which the current RCPT TO address corresponds,
1549 after a mapping was performed during transit. An ORCPT is the
1550 only reliable way to correlate a DSN from a multi-Recipient
1551 message transfer with the intended Recipient.
1553 RFC5321.Received: Set by - Originator, Relay, Mediator, Dest
1555 This field contains trace information, including originating host,
1556 Relays, Mediators, and MSA host domain names and/or IP Addresses.
1558 RFC5321.Return-Path: Set by - Originator
1560 The MDA records the RFC5321.MailFrom address into the
1561 RFC5321.Return-Path field.
1563 RFC2919.List-Id: Set by - Mediator, Author
1565 This field provides a globally unique Mailing List naming
1566 framework that is independent of particular hosts [RFC2919].
1570Crocker Informational [Page 28]
1572RFC 5598 Email Architecture July 2009
1575 The identifier is in the form of a domain name; however, the
1576 string usually is constructed by combining the two parts of an
1577 email address. The result is rarely a true domain name, listed in
1578 the domain name service, although it can be.
1580 RFC2369.List-*: Set by - Mediator, Author
1582 [RFC2369] defines a collection of message header fields for use by
1583 Mailing Lists. In effect, they supply list-specific parameters
1584 for common Mailing-List user operations. The identifiers for
1585 these operations are for the list itself and the user-as-
1586 subscriber [RFC2369].
1588 RFC0791.SourceAddr: Set by - The Client SMTP sending host
1589 immediately preceding the current receiving SMTP server
1591 [RFC0791] defines the basic unit of data transfer for the
1592 Internet: the IP datagram. It contains a Source Address field
1593 that specifies the IP Address for the host (interface) from which
1594 the datagram was sent. This information is set and provided by
1595 the IP layer, which makes it independent of mail-level mechanisms.
1596 As such, it is often taken to be authoritative, although it is
1597 possible to provide false addresses.
15994.2. User-Level Services
1601 Interactions at the user level entail protocol exchanges, distinct
1602 from those that occur at lower layers of the Internet Mail MHS
1603 architecture that is, in turn, above the Internet Transport layer.
1604 Because the motivation for email, and much of its use, is for
1605 interaction among people, the nature and details of these protocol
1606 exchanges often are determined by the needs of interpersonal and
1607 group communication. To accommodate the idiosyncratic behavior
1608 inherent in such communication, only subjective guidelines, rather
1609 than strict rules, can be offered for some aspects of system
1610 behavior. Mailing Lists provide particularly salient examples.
16124.2.1. Message User Agent (MUA)
1614 A Message User Agent (MUA) works on behalf of User Actors and User
1615 applications. It is their representative within the email service.
1617 The Author MUA (aMUA) creates a message and performs initial
1618 submission into the transfer infrastructure via a Mail Submission
1619 Agent (MSA). It can also perform any creation- and posting-time
1620 archiving in its Message Store (aMS). An MUA aMS can organize
1621 messages in many different ways. A common model uses aggregations,
1622 called "folders"; in IMAP they are called "mailboxes". This model
1626Crocker Informational [Page 29]
1628RFC 5598 Email Architecture July 2009
1631 allows a folder for messages under development (Drafts), a folder for
1632 messages waiting to be sent (Queued or Unsent), and a folder for
1633 messages that have been successfully posted for transfer (Sent). But
1634 none of these folders is required. For example, IMAP allows drafts
1635 to be stored in any folder, so no Drafts folder needs to be present.
1637 The Recipient MUA (rMUA) works on behalf of the Recipient to process
1638 received mail. This processing includes generating user-level
1639 disposition control messages, displaying and disposing of the
1640 received message, and closing or expanding the user-communication
1641 loop by initiating replies and forwarding new messages.
1643 NOTE: Although not shown in Figure 5, an MUA itself can have a
1644 distributed implementation, such as a "thin" user-interface
1645 module on a constrained device such as a smartphone, with
1646 most of the MUA functionality running remotely on a more
1647 capable server. An example of such an architecture might use
1648 IMAP [RFC3501] for most of the interactions between an MUA
1649 client and an MUA server. An approach for such scenarios is
1650 defined by [RFC4550].
1652 A Mediator is a special class of MUA. It performs message
1653 re-posting, as discussed in Section 2.1.
1655 An MUA can be automated, on behalf of a User who is not present at
1656 the time the MUA is active. One example is a bulk sending service
1657 that has a timed-initiation feature. These services are not to be
1658 confused with a Mailing List Mediator, since there is no incoming
1659 message triggering the activity of the automated service.
1661 A popular and problematic MUA is an automatic responder, such as one
1662 that sends out-of-office notices. This behavior might be confused
1663 with that of a Mediator, but this MUA is generating a new message.
1664 Automatic responders can annoy Users of Mailing Lists unless they
1667 The identity fields are relevant to a typical MUA:
1675 RFC5322.To, RFC5322.CC
1682Crocker Informational [Page 30]
1684RFC 5598 Email Architecture July 2009
16874.2.2. Message Store (MS)
1689 An MUA can employ a long-term Message Store (MS). Figure 5 depicts
1690 an Author's MS (aMS) and a Recipient's MS (rMS). An MS can be
1691 located on a remote server or on the same machine as the MUA.
1693 An MS acquires messages from an MDA either proactively by a local
1694 mechanism or even by a standardized mechanism such as SMTP(!), or
1695 reactively by using POP or IMAP. The MUA accesses the MS either by a
1696 local mechanism or by using POP or IMAP. Using POP for individual
1697 message accesses, rather than for bulk transfer, is relatively rare
17004.3. MHS-Level Services
17024.3.1. Mail Submission Agent (MSA)
1704 A Mail Submission Agent (MSA) accepts the message submitted by the
1705 aMUA and enforces the policies of the hosting ADMD and the
1706 requirements of Internet standards. An MSA represents an unusual
1707 functional dichotomy. It represents the interests of the Author
1708 (aMUA) during message posting, to facilitate posting success; it also
1709 represents the interests of the MHS. In the architecture, these
1710 responsibilities are modeled, as shown in Figure 5, by dividing the
1711 MSA into two sub-components, aMSA and hMSA, respectively. Transfer
1712 of responsibility for a single message, from an Author's environment
1713 to the MHS, is called "posting". In Figure 5, it is marked as the
1714 (S) transition, within the MSA.
1716 The hMSA takes transit responsibility for a message that conforms to
1717 the relevant Internet standards and to local site policies. It
1718 rejects messages that are not in conformance. The MSA performs final
1719 message preparation for submission and effects the transfer of
1720 responsibility to the MHS, via the hMSA. The amount of preparation
1721 depends upon the local implementations. Examples of aMSA tasks
1722 include adding header fields, such as Date: and Message-ID:, and
1723 modifying portions of the message from local notations to Internet
1724 standards, such as expanding an address to its formal IMF
1727 Historically, standards-based MUA/MSA message postings have used SMTP
1728 [RFC5321]. The standard currently preferred is SUBMISSION [RFC4409].
1729 Although SUBMISSION derives from SMTP, it uses a separate TCP port
1730 and imposes distinct requirements, such as access authorization.
1738Crocker Informational [Page 31]
1740RFC 5598 Email Architecture July 2009
1743 These identities are relevant to the MSA:
17574.3.2. Message Transfer Agent (MTA)
1759 A Message Transfer Agent (MTA) relays mail for one application-level
1760 "hop". It is like a packet switch or IP router in that its job is to
1761 make routing assessments and to move the message closer to the
1762 Recipients. Of course, email objects are typically much larger than
1763 the payload of a packet or datagram, and the end-to-end latencies are
1764 typically much higher. Relaying is performed by a sequence of MTAs
1765 until the message reaches a destination MDA. Hence, an MTA
1766 implements both client and server MTA functionality; it does not
1767 change addresses in the envelope or reformulate the editorial
1768 content. A change in data form, such as to MIME Content-Transfer-
1769 Encoding, is within the purview of an MTA, but removal or replacement
1770 of body content is not. An MTA also adds trace information
1773 NOTE: Within a destination ADMD, email-relaying modules can make a
1774 variety of changes to the message, prior to delivery. In
1775 such cases, these modules are acting as Gateways, rather than
1778 Internet Mail uses SMTP ([RFC5321], [RFC2821], [RFC0821]) primarily
1779 to effect point-to-point transfers between peer MTAs. Other transfer
1780 mechanisms include Batch SMTP [RFC2442] and On-Demand Mail Relay
1781 (ODMR) SMTP [RFC2645]. As with most network-layer mechanisms, the
1782 Internet Mail SMTP supports a basic level of reliability, by virtue
1783 of providing for retransmission after a temporary transfer failure.
1784 Unlike typical packet switches (and Instant Messaging services),
1785 Internet Mail MTAs are expected to store messages in a manner that
1786 allows recovery across service interruptions, such as host-system
1787 shutdown. The degree of such robustness and persistence by an MTA
1788 can vary. The base SMTP specification provides a framework for
1789 protocol response codes. An extensible enhancement to this framework
1790 is defined in [RFC5248].
1794Crocker Informational [Page 32]
1796RFC 5598 Email Architecture July 2009
1799 Although quite basic, the dominant routing mechanism for Internet
1800 Mail is the DNS MX record [RFC1035], which specifies an MTA through
1801 which the queried domain can be reached. This mechanism presumes a
1802 public, or at least a common, backbone that permits any attached MTA
1803 to connect to any other.
1805 MTAs can perform any of these well-established roles:
1807 Boundary MTA: An MTA that is part of an ADMD and interacts with MTAs
1808 in other ADMDs. This is also called a Border MTA.
1809 There can be different Boundary MTAs, according to the
1810 direction of mail-flow.
1812 Outbound MTA: An MTA that relays messages to other
1815 Inbound MTA: An MTA that receives inbound SMTP
1816 messages from MTA Relays in other
1817 ADMDs, for example, an MTA running on
1818 the host listed as the target of an MX
1821 Final MTA: The MTA that transfers a message to the MDA.
1823 These identities are relevant to the MTA:
1833 RFC5322.Received: Set by - Relay Server
18374.3.3. Mail Delivery Agent (MDA)
1839 A transfer of responsibility from the MHS to a Recipient's
1840 environment (mailbox) is called "delivery". In the architecture, as
1841 depicted in Figure 5, delivery takes place within a Mail Delivery
1842 Agent (MDA) and is shown as the (D) transition from the MHS-oriented
1843 MDA component (hMDA) to the Recipient-oriented MDA component (rMDA).
1850Crocker Informational [Page 33]
1852RFC 5598 Email Architecture July 2009
1855 An MDA can provide distinctive, address-based functionality, made
1856 possible by its detailed information about the properties of the
1857 destination address. This information might also be present
1858 elsewhere in the Recipient's ADMD, such as at an organizational
1859 border (Boundary) Relay. However, it is required for the MDA, if
1860 only because the MDA is required to know where to deliver the
1863 Like an MSA, an MDA serves two roles, as depicted in Figure 5.
1864 Formal transfer of responsibility, called "delivery", is effected
1865 between the two components that embody these roles and is shown as
1866 "(D)" in Figure 5. The MHS portion (hMDA) primarily functions as a
1867 server SMTP engine. A common additional role is to redirect the
1868 message to an alternative address, as specified by the Recipient
1869 addressee's preferences. The job of the Recipient portion of the MDA
1870 (rMDA) is to perform any delivery actions that the Recipient
1873 Transfer into the MDA is accomplished by a normal MTA transfer
1874 mechanism. Transfer from an MDA to an MS uses an access protocol,
1875 such as POP or IMAP.
1877 NOTE: The term "delivery" can refer to the formal, MHS function
1878 specified here or to the first time a message is displayed to
1879 a Recipient. A simple, practical test for whether the MHS-
1880 based definition applies is whether a DSN can be generated.
1882 These identities are relevant to the MDA:
1884 RFC5321.Return-Path: Set by - Author Originator or Mediator
1887 The MDA records the RFC5321.MailFrom address into the
1888 RFC5321.Return-Path field.
1890 RFC5322.Received: Set by - MDA server
1892 An MDA can record a Received: header field to indicate trace
1893 information, including source host and receiving host domain
1894 names and/or IP Addresses.
18964.4. Transition Modes
1898 From the origination site to the point of delivery, Internet Mail
1899 usually follows a "push" model. That is, the Actor that holds the
1900 message initiates transfer to the next venue, typically with SMTP
1901 [RFC5321] or the Local Mail Transfer Protocol (LMTP) [RFC2033]. With
1902 a "pull" model, the Actor that holds the message waits for the Actor
1906Crocker Informational [Page 34]
1908RFC 5598 Email Architecture July 2009
1911 in the next venue to initiate a request for transfer. Standardized
1912 mechanisms for pull-based MHS transfer are ETRN [RFC1985] and ODMR
1915 After delivery, the Recipient's MUA (or MS) can gain access by having
1916 the message pushed to it or by having the receiver of access pull the
1917 message, such as by using POP [RFC1939] and IMAP [RFC3501].
19194.5. Implementation and Operation
1921 A discussion of any interesting system architecture often bogs down
1922 when architecture and implementation are confused. An architecture
1923 defines the conceptual functions of a service, divided into discrete
1924 conceptual modules. An implementation of that architecture can
1925 combine or separate architectural components, as needed for a
1926 particular operational environment. For example, a software system
1927 that primarily performs message relaying is an MTA, yet it might also
1928 include MDA functionality. That same MTA system might be able to
1929 interface with non-Internet email services and thus perform both as
1930 an MTA and as a Gateway.
1932 Similarly, implemented modules might be configured to form
1933 elaborations of the architecture. An interesting example is a
1934 distributed MS. One portion might be a remote server and another
1935 might be local to the MUA. As discussed in [RFC1733], there are
1936 three operational relationships among such MSs:
1938 Online: The MS is remote, and messages are accessible only when the
1939 MUA is attached to the MS so that the MUA will re-fetch all or
1940 part of a message from one session to the next.
1942 Offline: The MS is local to the User, and messages are completely
1943 moved from any remote store, rather than (also) being retained
1946 Disconnected: An rMS and a uMS are kept synchronized, for all or
1947 part of their contents, while they are connected. When they are
1948 disconnected, mail can arrive at the rMS and the User can make
1949 changes to the uMS. The two stores are re-synchronized when they
1954 Basic message transfer from Author to Recipients is accomplished by
1955 using an asynchronous store-and-forward communication infrastructure
1956 in a sequence of independent transmissions through some number of
1957 MTAs. A very different task is a sequence of postings and deliveries
1958 through Mediators. A Mediator forwards a message through a
1962Crocker Informational [Page 35]
1964RFC 5598 Email Architecture July 2009
1967 re-posting process. The Mediator shares some functionality with
1968 basic MTA relaying, but has greater flexibility in both addressing
1969 and content than is available to MTAs.
1971 This is the core set of message information that is commonly set by
1972 all types of Mediators:
1974 RFC5321.HELO/.EHLO: Set by - Mediator Originator
1976 RFC3461.ENVID: Set by - Mediator Originator
1978 RFC5321.RcptTo: Set by - Mediator Author
1980 RFC5321.Received: Set by - Mediator Dest
1982 The Mediator can record received information to indicate the
1983 delivery to the original address and submission to the alias
1984 address. The trace of Received: header fields can include
1985 everything from original posting, through relaying, to final
1988 The aspect of a Mediator that distinguishes it from any other MUA
1989 creating a message is that a Mediator preserves the integrity and
1990 tone of the original message, including the essential aspects of its
1991 origination information. The Mediator might also add commentary.
1993 Examples of MUA messages that a Mediator does not create include:
1995 New message that forwards an existing message:
1997 Although this action provides a basic template for a class of
1998 Mediators, its typical occurrence is not, itself, an example of
1999 a Mediator. The new message is viewed as being from the Actor
2000 that is doing the forwarding, rather than from the original
2002 A new message encapsulates the original message and is seen as
2003 from the new Originator. This Mediator Originator might add
2004 commentary and can modify the original message content.
2005 Because the forwarded message is a component of the message
2006 sent by the new Originator, the new message creates a new
2007 dialogue. However, the final Recipient still sees the
2008 contained message as from the original Author.
2012 When a Recipient responds to the Author of a message, the new
2013 message is not typically viewed as a forwarding of the
2014 original. Its focus is the new content, although it might
2018Crocker Informational [Page 36]
2020RFC 5598 Email Architecture July 2009
2023 contain all or part of the material from the original message.
2024 The earlier material is merely contextual and secondary. This
2025 includes automated replies, such as vacation out-of-office
2026 notices, as discussed in Section 4.2.1.
2030 The integrity of the original message is usually preserved, but
2031 one or more comments about the message are added in a manner
2032 that distinguishes commentary from original text. The primary
2033 purpose of the new message is to provide commentary from a new
2034 Author, similar to a Reply.
2036 The remainder of this section describes common examples of Mediators.
2040 One function of an MDA is to determine the internal location of a
2041 mailbox in order to perform delivery. An Alias is a simple
2042 re-addressing facility that provides one or more new Internet Mail
2043 addresses, rather than a single, internal one; the message continues
2044 through the transfer service, for delivery to one or more alternate
2045 addresses. Although typically implemented as part of an MDA, this
2046 facility is a Recipient function. It resubmits the message, although
2047 all handling information except the envelope Recipient
2048 (rfc5321.RcptTo) address is retained. In particular, the Return
2049 Address (rfc5321.MailFrom) is unchanged.
2051 What is distinctive about this forwarding mechanism is how closely it
2052 resembles normal MTA store-and-forward relaying. Its only
2053 significant difference is that it changes the RFC5321.RcptTo value.
2054 Because this change is so small, aliasing can be viewed as a part of
2055 the lower-level mail-relaying activity. However, this small change
2056 has a large semantic impact: The designated Recipient has chosen a
2059 NOTE: When the replacement list includes more than one address, the
2060 alias is increasingly likely to have delivery problems. Any
2061 problem reports go to the original Author, not the
2062 administrator of the alias entry. This makes it more
2063 difficult to resolve the problem, because the original Author
2064 has no knowledge of the Alias mechanism.
2066 Including the core set of message information listed at the beginning
2067 of this section, Alias typically changes:
2074Crocker Informational [Page 37]
2076RFC 5598 Email Architecture July 2009
2079 RFC5322.To/.CC/.BCC: Set by - Author
2081 These fields retain their original addresses.
2083 RFC5321.MailFrom: Set by - Author
2085 The benefit of retaining the original MailFrom value is to
2086 ensure that an Actor related to the originating ADMD knows
2087 there has been a delivery problem. On the other hand, the
2088 responsibility for handling problems, when transiting from the
2089 original Recipient mailbox to the alias mailbox usually lies
2090 with that original Recipient, because the Alias mechanism is
2091 strictly under that Recipient's control. Retaining the
2092 original MailFrom address prevents this.
2096 Also called the ReDirector, the ReSender's actions differ from
2097 forwarding because the Mediator "splices" a message's addressing
2098 information to connect the Author of the original message with the
2099 Recipient of the new message. This connection permits them to have
2100 direct exchange, using their normal MUA Reply functions, while also
2101 recording full reference information about the Recipient who served
2102 as a Mediator. Hence, the new Recipient sees the message as being
2103 from the original Author, even if the Mediator adds commentary.
2105 Including the core set of message information listed at the beginning
2106 of this section, these identities are relevant to a resent message:
2108 RFC5322.From: Set by - original Author
2110 Names and addresses for the original Author of the message
2111 content are retained. The free-form (display-name) portion of
2112 the address might be modified to provide an informal reference
2115 RFC5322.Reply-To: Set by - original Author
2117 If this field is present in the original message, it is
2118 retained in the resent message.
2120 RFC5322.Sender: Set by - Author's Originator or Mediator
2123 RFC5322.To/.CC/.BCC: Set by - original Author
2125 These fields specify the original message Recipients.
2130Crocker Informational [Page 38]
2132RFC 5598 Email Architecture July 2009
2135 RFC5322.Resent-From: Set by - Mediator Author
2137 This address is of the original Recipient who is redirecting
2138 the message. Otherwise, the same rules apply to the Resent-
2139 From: field as to an original RFC5322.From field.
2141 RFC5322.Resent-Sender: Set by - Mediator Originator
2143 The address of the Actor responsible for resubmitting the
2144 message. As with RFC5322.Sender, this field can be omitted
2145 when it contains the same address as RFC5322.Resent-From.
2147 RFC5322.Resent-To/-CC/-BCC: Set by - Mediator Author
2149 The addresses of the new Recipients who are now able to reply
2150 to the original Author.
2152 RFC5321.MailFrom: Set by - Mediator Originator
2154 The Actor responsible for resubmission (RFC5322.Resent-Sender)
2155 is also responsible for specifying the new MailFrom address.
2159 A Mailing List receives messages as an explicit addressee and then
2160 re-posts them to a list of subscribed members. The Mailing List
2161 performs a task that can be viewed as an elaboration of the ReSender.
2162 In addition to sending the new message to a potentially large number
2163 of new Recipients, the Mailing List can modify content, for example,
2164 by deleting attachments, converting the format, and adding list-
2165 specific comments. Mailing Lists also archive messages posted by
2166 Authors. Still the message retains characteristics of being from the
2169 Including the core set of message information listed at the beginning
2170 of this section, these identities are relevant to a Mailing List
2171 processor when submitting a message:
2173 RFC2919.List-Id: Set by - Mediator Author
2175 RFC2369.List-*: Set by - Mediator Author
2177 RFC5322.From: Set by - original Author
2179 Names and email addresses for the original Author of the
2180 message content are retained.
2186Crocker Informational [Page 39]
2188RFC 5598 Email Architecture July 2009
2191 RFC5322.Reply-To: Set by - Mediator or original Author
2193 Although problematic, it is common for a Mailing List to assign
2194 its own addresses to the Reply-To: header field of messages
2195 that it posts. This assignment is intended to ensure that
2196 replies go to all list members, rather than to only the
2197 original Author. As a User Actor, a Mailing List is the Author
2198 of the new message and can legitimately set the Reply-To:
2199 value. As a Mediator attempting to represent the message on
2200 behalf of its original Author, creating or modifying a
2201 Reply-To: field can be viewed as violating that Author's
2202 intent. When the Reply-To is modified in this way, a reply
2203 that is meant only for the original Author will instead go to
2204 the entire list. When the Mailing List does not set the field,
2205 a reply meant for the entire list can instead go only to the
2206 original Author. At best, either choice is a matter of group
2207 culture for the particular list.
2209 RFC5322.Sender: Set by - Author Originator or Mediator Originator
2211 This field usually specifies the address of the Actor
2212 responsible for Mailing List operations. Mailing Lists that
2213 operate in a manner similar to a simple MTA Relay preserve as
2214 much of the original handling information as possible,
2215 including the original RFC5322.Sender field. (Note that this
2216 mode of operation causes the Mailing List to behave much like
2217 an Alias, with a possible difference in number of new
2220 RFC5322.To/.CC: Set by - original Author
2222 These fields usually contain the original list of Recipient
2225 RFC5321.MailFrom: Set by - Mediator Originator
2227 Because a Mailing List can modify the content of a message in
2228 any way, it is responsible for that content; that is, it is an
2229 Author. As such, the Return Address is specified by the
2230 Mailing List. Although it is plausible for the Mailing List to
2231 reuse the Return Address employed by the original Originator,
2232 notifications sent to that address after a message has been
2233 processed by a Mailing List could be problematic.
2242Crocker Informational [Page 40]
2244RFC 5598 Email Architecture July 2009
2249 A Gateway performs the basic routing and transfer work of message
2250 relaying, but it also is permitted to modify content, structure,
2251 address, or attributes as needed to send the message into a messaging
2252 environment that operates under different standards or potentially
2253 incompatible policies. When a Gateway connects two differing
2254 messaging services, its role is easy to identify and understand.
2255 When it connects environments that follow similar technical
2256 standards, but significantly different administrative policies, it is
2257 easy to view a Gateway as merely an MTA.
2259 The critical distinction between an MTA and a Gateway is that a
2260 Gateway can make substantive changes to a message to map between the
2261 standards. In virtually all cases, this mapping results in some
2262 degree of semantic loss. The challenge of Gateway design is to
2263 minimize this loss. Standardized Gateways to Internet Mail are
2264 facsimile [RFC4143], voicemail [RFC3801], and the Multimedia
2265 Messaging Service (MMS) [RFC4356].
2267 A Gateway can set any identity field available to an MUA. Including
2268 the core set of message information listed at the beginning of this
2269 section, these identities are typically relevant to Gateways:
2271 RFC5322.From: Set by - original Author
2273 Names and addresses for the original Author of the message
2274 content are retained. As for all original addressing
2275 information in the message, the Gateway can translate addresses
2276 as required to continue to be useful in the target environment.
2278 RFC5322.Reply-To: Set by - original Author
2280 It is best for a Gateway to retain this information, if it is
2281 present. The ability to perform a successful reply by a
2282 Recipient is a typical test of Gateway functionality.
2284 RFC5322.Sender: Set by - Author Originator or Mediator Originator
2286 This field can retain the original value or can be set to a new
2289 RFC5322.To/.CC/.BCC: Set by - original Recipient
2291 These fields usually retain their original addresses.
2298Crocker Informational [Page 41]
2300RFC 5598 Email Architecture July 2009
2303 RFC5321.MailFrom: Set by - Author Originator or Mediator
2306 The Actor responsible for handling the message can specify a
2307 new address to receive handling notices.
2311 To enforce security boundaries, organizations can subject messages to
2312 analysis for conformance with its safety policies. An example is
2313 detection of content classed as spam or a virus. A filter might
2314 alter the content to render it safe, such as by removing content
2315 deemed unacceptable. Typically, these actions add content to the
2316 message that records the actions.
23206.1. Security Considerations
2322 This document describes the existing Internet Mail architecture. It
2323 introduces no new capabilities. The security considerations of this
2324 deployed architecture are documented extensively in the technical
2325 specifications referenced by this document. These specifications
2326 cover classic security topics, such as authentication and privacy.
2327 For example, email-transfer protocols can use standardized mechanisms
2328 for operation over authenticated and/or encrypted links, and message
2329 content has similar protection standards available. Examples of such
2330 mechanisms include SMTP-TLS [RFC3207], SMTP-Auth [RFC4954], OpenPGP
2331 [RFC4880], and S/MIME [RFC3851].
2333 The core of the Internet Mail architecture does not impose any
2334 security requirements or functions on the end-to-end or hop-by-hop
2335 components. For example, it does not require participant
2336 authentication and does not attempt to prevent data disclosure.
2338 Particular message attributes might expose specific security
2339 considerations. For example, the blind carbon copy feature of the
2340 architecture invites disclosure concerns, as discussed in Section 7.2
2341 of [RFC5321] and Section 5 of [RFC5322]. Transport of text or non-
2342 text content in this architecture has security considerations that
2343 are discussed in [RFC5322], [RFC2045], [RFC2046], and [RFC4288];
2344 also, security considerations are present for some of the media types
2345 registered with IANA.
2347 Agents that automatically respond to email raise significant security
2348 considerations, as discussed in [RFC3834]. Gateway behaviors affect
2349 end-to-end security services, as discussed in [RFC2480]. Security
2350 considerations for boundary filters are discussed in [RFC5228].
2354Crocker Informational [Page 42]
2356RFC 5598 Email Architecture July 2009
2359 See Section 7.1 of [RFC5321] for a discussion of the topic of
2360 origination validation. As mentioned in Section 4.1.4, it is common
2361 practice for components of this architecture to use the
2362 RFC0791.SourceAddr to make policy decisions [RFC2505], although the
2363 address can be "spoofed". It is possible to use it without
2364 authorization. SMTP and Submission authentication ([RFC4409],
2365 [RFC4954]) provide more secure alternatives.
2367 The discussion of trust boundaries, ADMDs, Actors, roles, and
2368 responsibilities in this document highlights the relevance and
2369 potential complexity of security factors for operation of an Internet
2370 Mail service. The core design of Internet Mail to encourage open and
2371 casual exchange of messages has met with scaling challenges, as the
2372 population of email participants has grown to include those with
2373 problematic practices. For example, spam, as defined in [RFC2505],
2374 is a by-product of this architecture. A number of Standards Track or
2375 BCP documents on the subject have been issued (see [RFC2505],
2376 [RFC5068], and [RFC5235]).
23786.2. Internationalization
2380 The core Internet email standards are based on the use of US-ASCII --
2381 that is, SMTP [RFC5321] and IMF [RFC5322], as well as their
2382 predecessors. They describe the transport and composition of
2383 messages as composed strictly of US-ASCII 7-bit encoded characters.
2384 The standards have been incrementally enhanced to allow for
2385 characters outside of this limited set, while retaining mechanisms
2386 for backwards-compatibility. Specifically:
2388 o The MIME specifications ([RFC2045], [RFC2046], [RFC2047],
2389 [RFC2049]) allow for the use of coded character sets and
2390 character-encoding schemes ("charsets" in MIME terminology) other
2391 than US-ASCII. MIME's [RFC2046] allows the textual content of a
2392 message to have a label affixed that specifies the charset used in
2393 that content. Equally, MIME's [RFC2047] allows the textual
2394 content of certain header fields in a message to be similarly
2395 labeled. However, since messages might be transported over SMTP
2396 implementations only capable of transporting 7-bit encoded
2397 characters, MIME's [RFC2045] also provides for "content transfer
2398 encoding" so that characters of other charsets can be re-encoded
2399 as an overlay to US-ASCII.
2401 o MIME's [RFC2045] allows for the textual content of a message to be
2402 in an 8-bit character-encoding scheme. In order to transport
2403 these without re-encoding them, the SMTP specification supports an
2404 option [RFC1652] that permits the transport of such textual
2410Crocker Informational [Page 43]
2412RFC 5598 Email Architecture July 2009
2415 content. However, the [RFC1652] option does not address the use
2416 of 8-bit content in message header fields, and therefore [RFC2047]
2417 encoding is still required for those.
2419 o A series of experimental protocols on Email Address
2420 Internationalization (EAI) have been released that extend SMTP and
2421 IMF to allow for 8-bit encoded characters to appear in addresses
2422 and other information throughout the header fields of messages.
2423 [RFC5335] specifies the format of such message header fields
2424 (which encode the characters in UTF-8), and [RFC5336] specifies an
2425 SMTP option for the transport of these messages.
2427 o MIME's [RFC2045] and [RFC2046] allow for the transport of true
2428 multimedia material; such material enables internationalization
2429 because it is not restricted to any particular language or locale.
2431 o The formats for Delivery Status Notifications (DSNs -- [RFC3462],
2432 [RFC3463], [RFC3464]) and Message Disposition Notifications (MDNs
2433 -- [RFC3798]) include both a structured and unstructured
2434 representation of the notification. In the event that the
2435 unstructured representation is in the wrong language or is
2436 otherwise unsuitable for use, this allows an MUA to construct its
2437 own appropriately localized representation of notification for
2438 display to the User.
2440 o POP and IMAP have no difficulties with handling MIME messages,
2441 including ones containing 8bit, and therefore are not a source of
2442 internationalization issues.
2444 Hence, the use of UTF-8 is fully established in existing Internet
2445 Mail. However, support for long-standing encoding forms is retained
2466Crocker Informational [Page 44]
2468RFC 5598 Email Architecture July 2009
24737.1. Normative References
2475 [RFC0791] Postel, J., "Internet Protocol", STD 5, RFC 791,
2478 [RFC1034] Mockapetris, P., "Domain names - concepts and facilities",
2479 STD 13, RFC 1034, November 1987.
2481 [RFC1035] Mockapetris, P., "Domain names - implementation and
2482 specification", STD 13, RFC 1035, November 1987.
2484 [RFC1939] Myers, J. and M. Rose, "Post Office Protocol - Version 3",
2485 STD 53, RFC 1939, May 1996.
2487 [RFC2045] Freed, N. and N. Borenstein, "Multipurpose Internet Mail
2488 Extensions (MIME) Part One: Format of Internet Message
2489 Bodies", RFC 2045, November 1996.
2491 [RFC2046] Freed, N. and N. Borenstein, "Multipurpose Internet Mail
2492 Extensions (MIME) Part Two: Media Types", RFC 2046,
2495 [RFC2047] Moore, K., "MIME (Multipurpose Internet Mail Extensions)
2496 Part Three: Message Header Extensions for Non-ASCII Text",
2497 RFC 2047, November 1996.
2499 [RFC2049] Freed, N. and N. Borenstein, "Multipurpose Internet Mail
2500 Extensions (MIME) Part Five: Conformance Criteria and
2501 Examples", RFC 2049, November 1996.
2503 [RFC2181] Elz, R. and R. Bush, "Clarifications to the DNS
2504 Specification", RFC 2181, July 1997.
2506 [RFC2369] Neufeld, G. and J. Baer, "The Use of URLs as Meta-Syntax
2507 for Core Mail List Commands and their Transport through
2508 Message Header Fields", RFC 2369, July 1998.
2510 [RFC2645] Gellens, R., "ON-DEMAND MAIL RELAY (ODMR) SMTP with
2511 Dynamic IP Addresses", RFC 2645, August 1999.
2513 [RFC2919] Chandhok, R. and G. Wenger, "List-Id: A Structured Field
2514 and Namespace for the Identification of Mailing Lists",
2515 RFC 2919, March 2001.
2517 [RFC3192] Allocchio, C., "Minimal FAX address format in Internet
2518 Mail", RFC 3192, October 2001.
2522Crocker Informational [Page 45]
2524RFC 5598 Email Architecture July 2009
2527 [RFC3297] Klyne, G., Iwazaki, R., and D. Crocker, "Content
2528 Negotiation for Messaging Services based on Email",
2529 RFC 3297, July 2002.
2531 [RFC3458] Burger, E., Candell, E., Eliot, C., and G. Klyne, "Message
2532 Context for Internet Mail", RFC 3458, January 2003.
2534 [RFC3461] Moore, K., "Simple Mail Transfer Protocol (SMTP) Service
2535 Extension for Delivery Status Notifications (DSNs)",
2536 RFC 3461, January 2003.
2538 [RFC3462] Vaudreuil, G., "The Multipart/Report Content Type for the
2539 Reporting of Mail System Administrative Messages",
2540 RFC 3462, January 2003.
2542 [RFC3463] Vaudreuil, G., "Enhanced Mail System Status Codes",
2543 RFC 3463, January 2003.
2545 [RFC3501] Crispin, M., "INTERNET MESSAGE ACCESS PROTOCOL - VERSION
2546 4rev1", RFC 3501, March 2003.
2548 [RFC3798] Hansen, T. and G. Vaudreuil, "Message Disposition
2549 Notification", RFC 3798, May 2004.
2551 [RFC3834] Moore, K., "Recommendations for Automatic Responses to
2552 Electronic Mail", RFC 3834, August 2004.
2554 [RFC3864] Klyne, G., Nottingham, M., and J. Mogul, "Registration
2555 Procedures for Message Header Fields", BCP 90, RFC 3864,
2558 [RFC4021] Klyne, G. and J. Palme, "Registration of Mail and MIME
2559 Header Fields", RFC 4021, March 2005.
2561 [RFC4288] Freed, N. and J. Klensin, "Media Type Specifications and
2562 Registration Procedures", BCP 13, RFC 4288, December 2005.
2564 [RFC4289] Freed, N. and J. Klensin, "Multipurpose Internet Mail
2565 Extensions (MIME) Part Four: Registration Procedures",
2566 BCP 13, RFC 4289, December 2005.
2568 [RFC4409] Gellens, R. and J. Klensin, "Message Submission for Mail",
2569 RFC 4409, April 2006.
2571 [RFC4550] Maes, S. and A. Melnikov, "Internet Email to Support
2572 Diverse Service Environments (Lemonade) Profile",
2573 RFC 4550, June 2006.
2578Crocker Informational [Page 46]
2580RFC 5598 Email Architecture July 2009
2583 [RFC5228] Guenther, P. and T. Showalter, "Sieve: An Email Filtering
2584 Language", RFC 5228, January 2008.
2586 [RFC5248] Hansen, T. and J. Klensin, "A Registry for SMTP Enhanced
2587 Mail System Status Codes", BCP 138, RFC 5248, June 2008.
2589 [RFC5321] Klensin, J., "Simple Mail Transfer Protocol", RFC 5321,
2592 [RFC5322] Resnick, P., Ed., "Internet Message Format", RFC 5322,
25957.2. Informative References
2597 [RFC0733] Crocker, D., Vittal, J., Pogran, K., and D. Henderson,
2598 "Standard for the format of ARPA network text messages",
2599 RFC 733, November 1977.
2601 [RFC0821] Postel, J., "Simple Mail Transfer Protocol", STD 10,
2602 RFC 821, August 1982.
2604 [RFC0822] Crocker, D., "Standard for the format of ARPA Internet
2605 text messages", STD 11, RFC 822, August 1982.
2607 [RFC1506] Houttuin, J., "A Tutorial on Gatewaying between X.400 and
2608 Internet Mail", RFC 1506, August 1993.
2610 [RFC1652] Klensin, J., Freed, N., Rose, M., Stefferud, E., and D.
2611 Crocker, "SMTP Service Extension for 8bit-MIMEtransport",
2612 RFC 1652, July 1994.
2614 [RFC1733] Crispin, M., "Distributed Electronic Mail Models in
2615 IMAP4", RFC 1733, December 1994.
2617 [RFC1767] Crocker, D., "MIME Encapsulation of EDI Objects",
2618 RFC 1767, March 1995.
2620 [RFC1985] De Winter, J., "SMTP Service Extension for Remote Message
2621 Queue Starting", RFC 1985, August 1996.
2623 [RFC2033] Myers, J., "Local Mail Transfer Protocol", RFC 2033,
2626 [RFC2142] Crocker, D., "MAILBOX NAMES FOR COMMON SERVICES, ROLES AND
2627 FUNCTIONS", RFC 2142, May 1997.
2629 [RFC2442] Freed, N., Newman, D., and Hoy, M., "The Batch SMTP Media
2630 Type", RFC 2442, November 1998.
2634Crocker Informational [Page 47]
2636RFC 5598 Email Architecture July 2009
2639 [RFC2480] Freed, N., "Gateways and MIME Security Multiparts",
2640 RFC 2480, January 1999.
2642 [RFC2505] Lindberg, G., "Anti-Spam Recommendations for SMTP MTAs",
2643 BCP 30, RFC 2505, February 1999.
2645 [RFC2821] Klensin, J., "Simple Mail Transfer Protocol", RFC 2821,
2648 [RFC2822] Resnick, P., "Internet Message Format", RFC 2822,
2651 [RFC3207] Hoffman, P., "SMTP Service Extension for Secure SMTP over
2652 Transport Layer Security", RFC 3207, February 2002.
2654 [RFC3464] Moore, K. and G. Vaudreuil, "An Extensible Message Format
2655 for Delivery Status Notifications", RFC 3464,
2658 [RFC3801] Vaudreuil, G. and G. Parsons, "Voice Profile for Internet
2659 Mail - version 2 (VPIMv2)", RFC 3801, June 2004.
2661 [RFC3851] Ramsdell, B., "Secure/Multipurpose Internet Mail
2662 Extensions (S/MIME) Version 3.1 Message Specification",
2663 RFC 3851, July 2004.
2665 [RFC3885] Allman, E. and T. Hansen, "SMTP Service Extension for
2666 Message Tracking", RFC 3885, September 2004.
2668 [RFC4142] Crocker, D. and G. Klyne, "Full-mode Fax Profile for
2669 Internet Mail (FFPIM)", RFC 4142, November 2005.
2671 [RFC4143] Toyoda, K. and D. Crocker, "Facsimile Using Internet Mail
2672 (IFAX) Service of ENUM", RFC 4143, November 2005.
2674 [RFC4356] Gellens, R., "Mapping Between the Multimedia Messaging
2675 Service (MMS) and Internet Mail", RFC 4356, January 2006.
2677 [RFC4880] Callas, J., Donnerhacke, L., Finney, H., Shaw, D., and R.
2678 Thayer, "OpenPGP Message Format", RFC 4880, November 2007.
2680 [RFC4954] Siemborski, R. and A. Melnikov, "SMTP Service Extension
2681 for Authentication", RFC 4954, July 2007.
2683 [RFC5068] Hutzler, C., Crocker, D., Resnick, P., Allman, E., and T.
2684 Finch, "Email Submission Operations: Access and
2685 Accountability Requirements", BCP 134, RFC 5068,
2690Crocker Informational [Page 48]
2692RFC 5598 Email Architecture July 2009
2695 [RFC5235] Daboo, C., "Sieve Email Filtering: Spamtest and Virustest
2696 Extensions", RFC 5235, January 2008.
2698 [RFC5335] Abel, Y., "Internationalized Email Headers", RFC 5335,
2701 [RFC5336] Yao, J. and W. Mao, "SMTP Extension for Internationalized
2702 Email Addresses", RFC 5336, September 2008.
2704 [Tussle] Clark, D., Wroclawski, J., Sollins, K., and R. Braden,
2705 "Tussle in Cyberspace: Defining Tomorrow's Internet",
2746Crocker Informational [Page 49]
2748RFC 5598 Email Architecture July 2009
2751Appendix A. Acknowledgments
2753 This work began in 2004 and has evolved through numerous rounds of
2754 community review; it derives from a section in an early version of
2755 [RFC5068]. Over its 5 years of development, the document has gone
2756 through 14 incremental versions, with vigorous community review that
2757 produced many substantive changes. Review was performed in the IETF
2758 and other email technical venues. Although not a formal activity of
2759 the IETF, issues with the document's contents were resolved using the
2760 classic style of IETF community open, group decision-making. The
2761 document is already cited in other work, such as in IMAP and Sieve
2762 specifications and in academic classwork. The step of standardizing
2763 is useful to provide a solid and stable reference to the Internet's
2764 now-complex email service.
2766 Details of the Originator Actor role was greatly clarified during
2767 discussions in the IETF's Marid working group.
2769 Graham Klyne, Pete Resnick, and Steve Atkins provided thoughtful
2770 insight on the framework and details of the original drafts, as did
2771 Chris Newman for the final versions, while also serving as cognizant
2772 Area Director for the document. Tony Hansen served as document
2773 shepherd through the IETF process.
2775 Later reviews and suggestions were provided by Eric Allman, Nathaniel
2776 Borenstein, Ed Bradford, Cyrus Daboo, Frank Ellermann, Tony Finch,
2777 Ned Freed, Eric Hall, Willemien Hoogendoorn, Brad Knowles, John
2778 Leslie, Bruce Valdis Kletnieks, Mark E. Mallett, David MacQuigg,
2779 Alexey Melnikov, der Mouse, S. Moonesamy, Daryl Odnert, Rahmat M.
2780 Samik-Ibrahim, Marshall Rose, Hector Santos, Jochen Topf, Greg
2781 Vaudreuil, Patrick Cain, Paul Hoffman, Vijay Gurbani, and Hans
2784 Diligent early proof-reading was performed by Bruce Lilly. Diligent
2785 professional technical editing was provided by Susan Hunziker.
2787 The final stages of development for this document were guided by a
2788 design team comprising Alexey Melnikov, Pete Resnick, Carl S.
2789 Gutekunst, Jeff Macdonald, Randall Gellens, Tony Hansen, and Tony
2790 Finch. Pete Resnick developed the final version of the section on
2791 internationalization.
2802Crocker Informational [Page 50]
2804RFC 5598 Email Architecture July 2009
2825 actor 7, 19, 26, 28-29, 35-36, 38-40, 42-43, 49
2832 ADMD 12, 14-15, 19, 25, 31, 37
2833 Administrative Actors 14
2834 Administrative Management Domain 12
2847 content 11, 13-14, 20, 24, 32
2850 delivery 4, 10-11, 13-14, 18, 24-25, 35, 37-38
2851 Discussion of document 7
2852 domain name 17, 21, 28
2858Crocker Informational [Page 51]
2860RFC 5598 Email Architecture July 2009
2867 end-to-end 4-6, 11, 15, 28
2869 envelope 10, 13, 21, 24-25, 32, 37
2874 gateway 6, 12-13, 18, 25, 32
2881 identifier 18-19, 21, 25, 29
2882 IMAP 24, 31, 34-35, 44
2894 Mail Submission Agent 12
2895 mailbox 17, 19, 24, 28, 30, 33, 37-38
2899 Message Disposition Notification 10
2900 Message Handling Service 4
2901 Message Handling System 11
2902 Message Transfer Agent 4
2903 Message User Agent 4
2904 MHS 4, 10-13, 21-22, 24-25
2914Crocker Informational [Page 52]
2916RFC 5598 Email Architecture July 2009
2919 MUA 4, 14, 24, 30-31
2923 operations 3, 15, 18, 29, 40
2927 POP 24, 31, 34-35, 44
2928 posting 4, 10, 12, 21, 30-31, 35, 37
2970Crocker Informational [Page 53]
2972RFC 5598 Email Architecture July 2009
2978 Brandenburg InternetWorking
2983 Phone: +1.408.246.8253
2984 EMail: dcrocker@bbiw.net
3026Crocker Informational [Page 54]