Trivial File Transfer Protocol (TFTP)
The Internet protocols include a second file transfer server known as TFTP, the Trivial File Transfer Protocol. TFTP differs from FTP in several ways. First, the communication between a TFTP client and server uses user datagram protocol (UDP) instead of TCP. Second, TFTP supports only file transfer. That is, TFTP does not have a large set of commands. Most important, TFTP does not permit a user to list the contents of a discovery or interrogate the server to determine the names of files that are available. Third, TFTP does not have authorization. A client does not send a login name or password; a file can be transferred only if the file permission allows global access.
Although TFTP is less powerful than FTP, it has two marked advantages. First, TFTP can be used in environments where UDP is available, but TCP cannot be. Second, the code for TFTP requires less memory than the code for FTP. Although these advantages are not important in a general-purpose computer, they can be important in a small computer or a special-purpose hardware device.
TFTP is especially useful for bootstrapping a hardware device that does not have a disk on which to store system software. All the device needs is a network connection and a small amount of Read-Only Memory (ROM) into which TFTP, UDP and IP are hardwired. When it receives power, the device executes the code in ROM, which broadcasts a TFTP request across the network. A TFTP server on the network is configured to answer the request by sending a file that contains the binary program to be run. The device receives the file, loads it into memory, and begins to execute the program.
Bootstrapping over a network adds flexibility and reduces cost, because a separate server exists for each network, a server can be configured to supply a version of the software that is configured for the network. Cost is reduced because software can be changed without changing the hardware. For example, the manufacturer can release a new version of software for the device without changing the hardware or installing a new ROM.
ELECTRONIC MAIL
One of the most popular services is electronic mail (e-mail). Originally, electronic mail was designed as a straightforward extension of the traditional office memo. That is, the original e-mail systems were built to allow a person to communicate with other people; an individual created a message and specified other individuals as recipients. The e-mail software transmitted a copy of the message to each recipient.
Electronic mail systems have evolved from the original design, and are automated to permit more complex interactions. In particular, because a computer program can answer an e-mail message and send a reply, e-mail can be used in a variety of ways. For example, a company can establish a computer program that responds automatically to requests for information that arrive in mail messages. A user sends an e-mail request to the program, and receives the desired information in a reply.
Electronic Mailboxes and Addresses
An e-mail system uses many of the terms and concepts from a traditional office environment. Before e-mail can be sent to an individual, the person must be assigned an electronic mailbox. The mailbox consists of a passive storage area (say, a file on disk). Like a conventional mailbox, an electronic mailbox is private-the permissions are set to allow the mail software to add an incoming message to an arbitrary mailbox, but to deny anyone except the owner the right to examine or remove messages. In most cases, an electronic mailbox is associated with a computer account. Thus, a person who has multiple computer accounts can have multiple mailboxes.
Each electronic mailbox is assigned a unique electronic mail address (e-mail address); when someone sends a memo, they use an electronic mail address to specify a recipient. A full e-mail address contains two parts; the second specifies a computer and the first specifies a mailbox on that computer. The mailbox is a string that denotes a user's mailbox, and computer is a string that denotes the computer on which the mailbox is located (that is, a domain name). In the most widely used format, an 'at sign" separates the two components
mailbox@computer
E-mail software on the sender's computer uses the second part to determine which computer to contact and e-mail software on the recipient's computer uses the first part of the address to select a particular mailbox into which the message should be placed. Thus, the first part of an address is interpreted locally; the string may not have meaning outside of a single computer system.
Mail Transfer
After a user composes an e-mail message and specifies recipients, e-mail software transfers a copy of the message to each recipient. In most systems, two separate pieces of software are required. A user interacts with an e-mail interface program when composing or reading messages. The underlying e-mail system contains a mail transfer program that handles the details of sending a copy of a message to remote computer.When a user finishes composing an outgoing message, the e-mail interface places the message in a queue that the mail transfer program handles.
The mail transfer program waits for a message to be placed on its queue, and then transfers a copy of the message to each recipient. Sending a copy of a message to a recipient on the local computer is trivial because the transfer program can append the message to the user's mailbox. Sending a copy to a remote user is more complex. The mail transfer program becomes a client that contacts a server on the remote machine. The client sends the message to the server, which places a copy of the message in the recipient's mailbox. Figure 12.6 illustrates the interaction. Note that in the figure the mail transfer program on the sender's computer becomes a client of the remote mail server.
BACK | 
