Message Submission and Transfer

SMTP is the Internet's most venerable application-layer protocol. It lacks both the negotiation clumsiness of the Telnet protocol and the multichannel fumbling of the file transfer protocol. Its simplicity and elegance often lead to robust operation in multivendor environments. Its design also heavily inspired another now-ubiquitous Internet e-mail standard, Post Office Protocol (POP).

In order to understand SMTP, there are really only three basic concepts: the "where" of message envelopes, which describe the destination mailboxes; the "how" of address expansion, which describe the process whereby mailboxes are determined; and the "why" of protocol interactions, which describe the rules wherein responsibility for a message is transferred as it transits the Internet messaging infrastructure.

Message Envelopes

Because SMTP is a relaying protocol, it carries both the envelope and content of a message. The SMTP envelope is very simple. It contains:

• The e-mail address of the use that caused the creation of this envelope termed the originator address. This corresponds to the Sender: field in the headers, if present.
• One or more recipient e-mail addresses.

It must be strongly emphasized that there need be no relationship between the addresses in the SMTP envelope and any recipient addresses present in the headers of the message. In fact, it is entirely possible for the headers to contain no recipient addresses at all.⁴

Of historical interest, we note that SMTP also caries a "delivery mode" in the envelope, indicating how the message should be delivered, either to a mailbox or the recipient's "terminal," or both. In the modern Internet, only mailbox delivery is supported.

4 For example, as discussed in Chapter 2, the message might have an empty To: header or an empty Bcc: header.