Simple Mail Transfer Protocol

The Simple Mail Transfer Protocol (SMTP) is an Internet standard communication protocol for electronic mail transmission. Mail servers and other message transfer agents use SMTP to send and receive mail messages. User-level email clients typically use SMTP only for sending messages to a mail server for relaying, and typically submit outgoing email to the mail server on port 465 or 587 per RFC 8314. For retrieving messages, IMAP (which replaced the older POP3) is standard, but proprietary servers also often implement proprietary protocols, e.g., Exchange ActiveSync.

SMTP's origins began in 1980, building on concepts implemented on the ARPANET since 1971. It has been updated, modified and extended multiple times. The protocol version in common use today has extensible structure with various extensions for authentication, encryption, binary data transfer, and internationalized email addresses. SMTP servers commonly use the Transmission Control Protocol on port number 25 (between servers) and 587 (for submission from authenticated clients), both with or without encryption, and 465 with encryption for submission.

History

Predecessors to SMTP

Various forms of one-to-one electronic messaging were used in the 1960s. Users communicated using systems developed for specific mainframe computers. As more computers were interconnected, especially in the U.S. Government's ARPANET, standards were developed to permit exchange of messages between different operating systems.

Mail on the ARPANET traces its roots to 1971: the Mail Box Protocol, which was not implemented, but is discussed in RFC 196; and the SNDMSG program, which Ray Tomlinson of BBN adapted that year to send messages across two computers on the ARPANET. A further proposal for a Mail Protocol was made in RFC 524 in June 1973, which was not implemented.

The use of the File Transfer Protocol (FTP) for "network mail" on the ARPANET was proposed in RFC 469 in March 1973. Through RFC 561, RFC 680, RFC 724, and finally RFC 733 in November 1977, a standardized framework for "electronic mail" using FTP mail servers on was developed.

SMTP grew out of these standards developed during the 1970s. Ray Tomlinson discussed network mail among the International Network Working Group in INWG Protocol note 2, written in September 1974. INWG discussed protocols for electronic mail in 1979, which was referenced by Jon Postel in his early work on Internet email. Postel first proposed an Internet Message Protocol in 1979 as part of the Internet Experiment Note (IEN) series.

Original SMTP

In 1980, Postel and Suzanne Sluizer published RFC 772 which proposed the Mail Transfer Protocol as a replacement for the use of the FTP for mail. RFC 780 of May 1981 removed all references to FTP and allocated port 57 for TCP and UDP, an allocation that has since been removed by IANA. In November 1981, Postel published RFC 788 "Simple Mail Transfer Protocol".

The SMTP standard was developed around the same time as Usenet, a one-to-many communication network with some similarities.

SMTP became widely used in the early 1980s. At the time, it was a complement to the Unix to Unix Copy Program (UUCP), which was better suited for handling email transfers between machines that were intermittently connected. SMTP, on the other hand, works best when both the sending and receiving machines are connected to the network all the time. Both used a store and forward mechanism and are examples of push technology. Though Usenet's newsgroups were still propagated with UUCP between servers, UUCP as a mail transport has virtually disappeared along with the "bang paths" it used as message routing headers.

Sendmail, released with 4.1cBSD in 1983, was one of the first mail transfer agents (MTA) to implement SMTP. Over time, as BSD Unix became the most popular operating system on the Internet, Sendmail became the most common mail transfer agent.

The original SMTP protocol supported only unauthenticated unencrypted 7-bit ASCII text communications, susceptible to trivial man-in-the-middle attack, spoofing, and spamming, and requiring any binary data to be encoded to readable text before transmission. Due to absence of a proper authentication mechanism, by design every SMTP server was an open mail relay. The Internet Mail Consortium (IMC) reported that 55% of mail servers were open relays in 1998, but less than 1% in 2002. Because of spam concerns most email providers blocklist open relays, making original SMTP essentially impractical for general use on the Internet.

Modern SMTP

In November 1995, RFC 1869 defined Extended Simple Mail Transfer Protocol (ESMTP), which established a general structure for all existing and future extensions which aimed to add-in the features missing from the original SMTP. ESMTP defines consistent and manageable means by which ESMTP clients and servers can be identified and servers can indicate supported extensions.

Message submission (RFC 2476) and SMTP-AUTH (RFC 2554) were introduced in 1998 and 1999, both describing new trends in email delivery. Originally, SMTP servers were typically internal to an organization, receiving mail for the organization from the outside, and relaying messages from the organization to the outside. But as time went on, SMTP servers (mail transfer agents), in practice, were expanding their roles to become message submission agents for mail user agents, some of which were now relaying mail from the outside of an organization. (e.g. a company executive wishes to send email while on a trip using the corporate SMTP server.) This issue, a consequence of the rapid expansion and popularity of the World Wide Web, meant that SMTP had to include specific rules and methods for relaying mail and authenticating users to prevent abuses such as relaying of unsolicited email (spam). Work on message submission (RFC 2476) was originally started because popular mail servers would often rewrite mail in an attempt to fix problems in it, for example, adding a domain name to an unqualified address. This behavior is helpful when the message being fixed is an initial submission, but dangerous and harmful when the message originated elsewhere and is being relayed. Cleanly separating mail into submission and relay was seen as a way to permit and encourage rewriting submissions while prohibiting rewriting relay. As spam became more prevalent, it was also seen as a way to provide authorization for mail being sent out from an organization, as well as traceability. This separation of relay and submission quickly became a foundation for modern email security practices.

As this protocol started out purely ASCII text-based, it did not deal well with binary files, or characters in many non-English languages. Standards such as Multipurpose Internet Mail Extensions (MIME) were developed to encode binary files for transfer through SMTP. Mail transfer agents (MTAs) developed after Sendmail also tended to be implemented 8-bit clean, so that the alternate "just send eight" strategy could be used to transmit arbitrary text data (in any 8-bit ASCII-like character encoding) via SMTP. Mojibake was still a problem due to differing character set mappings between vendors, although the email addresses themselves still allowed only ASCII. 8-bit-clean MTAs today tend to support the 8BITMIME extension, permitting some binary files to be transmitted almost as easily as plain text (limits on line length and permitted octet values still apply, so that MIME encoding is needed for most non-text data and some text formats). In 2012, the SMTPUTF8 extension was created to support UTF-8 text, allowing international content and addresses in non-Latin scripts like Cyrillic or Chinese.

Many people contributed to the core SMTP specifications, among them Jon Postel, Eric Allman, Dave Crocker, Ned Freed, Randall Gellens, John Klensin, and Keith Moore.

Mail processing model

Email is submitted by a mail client (mail user agent, MUA) to a mail server (mail submission agent, MSA) using SMTP on TCP port 465 or 587. Most mailbox providers still allow submission on traditional port 25. The MSA delivers the mail to its mail transfer agent (MTA). Often, these two agents are instances of the same software launched with different options on the same machine. Local processing can be done either on a single machine, or split among multiple machines; mail agent processes on one machine can share files, but if processing is on multiple machines, they transfer messages between each other using SMTP, where each machine is configured to use the next machine as a smart host. Each process is an MTA (an SMTP server) in its own right.

The boundary MTA uses DNS to look up the MX (mail exchanger) record for the recipient's domain (the part of the email address on the right of @). The MX record contains the name of the target MTA. Based on the target host and other factors, the sending MTA selects a recipient server and connects to it to complete the mail exchange.

Message transfer can occur in a single connection between two MTAs, or in a series of hops through intermediary systems. A receiving SMTP server may be the ultimate destination, an intermediate "relay" (that is, it stores and forwards the message) or a "gateway" (that is, it may forward the message using some protocol other than SMTP). Per RFC 5321 section 2.1, each hop is a formal handoff of responsibility for the message, whereby the receiving server must either deliver the message or properly report the failure to do so.