Zen and the Art of Internet - Brendan P. Kehoe (good books to read for women txt) 📗

was born, freeing thousands of host numbers.

Each piece of an Internet address (like 192) is called an ``octet,'' representing one of four sets of eight bits. The first two or three pieces (e.g. 192.55.239) represent the network that a system is on, called its subnet. For example, all of the computers for Wesleyan University are in the subnet 129.133. They can have numbers like 129.133.10.10, 129.133.230.19, up to 65 thousand possible combinations (possible computers).

IP addresses and domain names aren't assigned arbitrarily---that would lead to unbelievable confusion. An application must be filed with the Network Information Center (NIC), either electronically (to hostmaster@nic.ddn.mil) or via regular mail.

Resolving Names and Numbers

Ok, computers can be referred to by either their FQDN or their Internet address. How can one user be expected to remember them all?

They aren't. The Internet is designed so that one can use either method. Since humans find it much more natural to deal with words than numbers in most cases, the FQDN for each host is mapped to its Internet number. Each domain is served by a computer within that domain, which provides all of the necessary information to go from a domain name to an IP address, and vice-versa. For example, when someone refers to foosun.bar.com, the resolver knows that it should ask the system foovax.bar.com about systems in bar.com. It asks what Internet address foosun.bar.com has; if the name foosun.bar.com really exists, foovax will send back its number. All of this ``magic'' happens behind the scenes.

Rarely will a user have to remember the Internet number of a site (although often you'll catch yourself remembering an apparently obscure number, simply because you've accessed the system frequently). However, you will remember a substantial number of FQDNs. It will eventually reach a point when you are able to make a reasonably accurate guess at what domain name a certain college, university, or company might have, given just their name.

The Networks

Internet The Internet is a large network of networks.'' There is no one network known as The Internet; rather, regional nets like SuraNet, PrepNet, NearNet, et al., are all inter-connected (nay,inter-networked'') together into one great living thing, communicating at amazing speeds with the TCP/IP protocol. All activity takes place in ``real-time.''

UUCP The UUCP network is a loose association of systems all communicating with the UUCP protocol. (UUCP stands for `Unix-to-Unix Copy Program'.) It's based on two systems connecting to each other at specified intervals, called polling, and executing any work scheduled for either of them. Historically most UUCP was done with Unix equipment, although the software's since been implemented on other platforms (e.g. VMS). For example, the system oregano polls the system basil once every two hours. If there's any mail waiting for oregano, basil will send it at that time; likewise, oregano will at that time send any jobs waiting for basil.

BITNET BITNET (the ``Because It's Time Network'') is comprised of systems connected by point-to-point links, all running the NJE protocol. It's continued to grow, but has found itself suffering at the hands of the falling costs of Internet connections. Also, a number of mail gateways are in place to reach users on other networks.

The Physical Connection

The actual connections between the various networks take a variety of forms. The most prevalent for Internet links are 56k leased lines (dedicated telephone lines carrying 56kilobit-per-second connections) and T1 links (special phone lines with 1Mbps connections). Also installed are T3 links, acting as backbones between major locations to carry a massive 45Mbps load of traffic.

These links are paid for by each institution to a local carrier (for example, Bell Atlantic owns PrepNet, the main provider in Pennsylvania). Also available are SLIP connections, which carry Internet traffic (packets) over high-speed modems.

UUCP links are made with modems (for the most part), that run from 1200 baud all the way up to as high as 38.4Kbps. As was mentioned in The Networks, the connections are of the store-and-forward variety. Also in use are Internet-based UUCP links (as if things weren't already confusing enough!). The systems do their UUCP traffic over TCP/IP connections, which give the UUCP-based network some blindingly fast ``hops,'' resulting in better connectivity for the network as a whole. UUCP connections first became popular in the 1970's, and have remained in wide-spread use ever since. Only with UUCP can Joe Smith correspond with someone across the country or around the world, for the price of a local telephone call.

BITNET links mostly take the form of 9600bps modems connected from site to site. Often places have three or more links going; the majority, however, look to ``upstream'' sites for their sole link to the network.

``The Glory and the Nothing of a Name'' Byron, {Churchill's Grave}

Electronic Mail

The desire to communicate is the essence of networking. People have always wanted to correspond with each other in the fastest way possible, short of normal conversation. Electronic mail (or email) is the most prevalent application of this in computer networking. It allows people to write back and forth without having to spend much time worrying about how the message actually gets delivered. As technology grows closer and closer to being a common part of daily life, the need to understand the many ways it can be utilized and how it works, at least to some level, is vital. part of daily life (as has been evidenced by the ISDN effort, the need to understand the many ways it can be utilized and how it works, at least to some level, is vital.

Email Addresses

Electronic mail is hinged around the concept of an address; the section on Networking Basics made some reference to it while introducing domains. Your email address provides all of the information required to get a message to you from anywhere in the world. An address doesn't necessarily have to go to a human being. It could be an archive server, {See Archive Servers, for a description.} a list of people, or even someone's pocket pager. These cases are the exception to the norm---mail to most addresses is read by human beings.

%@!.: Symbolic Cacophony

Email addresses usually appear in one of two forms---using the Internet format which contains @, an at''-sign, or using the UUCP format which contains !, an exclamation point, also called abang.'' The latter of the two, UUCP ``bang'' paths, is more restrictive, yet more clearly dictates how the mail will travel.

To reach Jim Morrison on the system south.america.org, one would address the mail as jm@south.america.org. But if Jim's account was on a UUCP site named brazil, then his address would be brazil!jm. If it's possible (and one exists), try to use the Internet form of an address; bang paths can fail if an intermediate site in the path happens to be down. There is a growing trend for UUCP sites to register Internet domain names, to help alleviate the problem of path failures.

Another symbol that enters the fray is %---it acts as an extra ``routing'' method. For example, if the UUCP site dream is connected to south.america.org, but doesn't have an Internet domain name of its own, a user debbie on dream can be reached by writing to the address not smallexample!

debbie%dream@south.america.org

The form is significant. This address says that the local system should first send the mail to south.america.org. There the address debbie%dream will turn into debbie@dream, which will hopefully be a valid address. Then south.america.org will handle getting the mail to the host dream, where it will be delivered locally to debbie.

All of the intricacies of email addressing methods are fully covered in the book ``!%@@:: A Directory of Electronic Mail Addressing and Networks'' published by O'Reilly and Associates, as part of their Nutshell Handbook series. It is a must for any active email user. Write to nuts@ora.com for ordering information.

Sending and Receiving Mail

We'll make one quick diversion from being OS-neuter here, to show you what it will look like to send and receive a mail message on a Unix system. Check with your system administrator for specific instructions related to mail at your site.

A person sending the author mail would probably do something like this:

% mail brendan@cs.widener.edu Subject: print job's stuck

I typed `print babe.gif' and it didn't work! Why??

The next time the author checked his mail, he would see it listed in his mailbox as:

% mail "/usr/spool/mail/brendan": 1 messages 1 new 1 unread U 1 joeuser@foo.widene Tue May 5 20:36 29/956 print job's stuck ?

which gives information on the sender of the email, when it was sent, and the subject of the message. He would probably use the reply command of Unix mail to send this response:

? r To: joeuser@@foo.widener.edu Subject: Re: print job's stuck

You shouldn't print binary files like GIFs to a printer!

Brendan

Try sending yourself mail a few times, to get used to your system's mailer. It'll save a lot of wasted aspirin for both you and your system administrator.

Anatomy of a Mail Header

An electronic mail message has a specific structure to it that's common across every type of computer system. {The standard is written down in RFC-822. See also RFCs for more info on how to get copies of the various RFCs.} A sample would be:

From bush@hq.mil Sat May 25 17:06:01 1991 Received: from hq.mil by house.gov with SMTP id AA21901 (4.1/SMI for dan@house.gov); Sat, 25 May 91 17:05:56 -0400 Date: Sat, 25 May 91 17:05:56 -0400 From: The President bush@hq.mil Message-Id: 9105252105.AA06631@hq.mil To: dan@senate.gov Subject: Meeting

Hi Dan .. we have a meeting at 9:30 a.m. with the Joint Chiefs. Please don't oversleep this time.

The first line, with From and the two lines for Received: are usually not very interesting. They give the real'' address that the mail is coming from (as opposed to the address you should reply to, which may look much different), and what places the mail went through to get to you. Over the Internet, there is always at least one Received: header and usually no more than four or five. When a message is sent using UUCP, one Received: header is added for each system that the mail passes through. This can often result in more than a dozen Received: headers. While they help with dissecting problems in mail delivery, odds are the average user will never want to see them. Most mail programs will filter out this kind ofcruft'' in a header.

The Date: header contains the date and time the message was sent. Likewise, the good'' address (as opposed toreal'' address) is laid out in the From: header. Sometimes it won't include the full name of the person (in this case The President), and may look different, but it should always contain an email address of some form.

The Message-ID: of a message is intended mainly for tracing mail routing, and is rarely of interest to normal users. Every Message-ID: is guaranteed to be unique.

To: lists the email address (or addresses) of the recipients of the message. There may be a Cc: header, listing additional addresses. Finally, a brief subject for the message goes in the Subject: header.

The exact order of a message's headers may vary from system to system, but it will always include these fundamental headers that