Worldnet: Getting Closer Every Day (September, 1987) ---------------------------------------------------- By Hank@Taunivm.Bitnet First off, let me say that I am on the other side of the fence. My job is to make sure the system I work for is secure and that there are no hackers or crackers trying to do damage to the system I am employed to defend. In one instance, I assisted the police in collecting all the necessary information to create a court case against a cracker. The kid in question (a high school student) ended up getting a year of civil work. I subscribe to this magazine not to learn how to do something illegal but rather to learn what others are trying to do to me. Knowledge is a tool and by hiding a tool you gain nothing. Therefore, I have decided to explain how international computer networks work, how they are tied together and what services you can hope to receive from them. There are dozens of computer networks - all of them spawning off the grandfather of all networks: ARPANET. Today, it has grown so large that it is known as The Internet. As more and more networks begin to interconnect, the concept of a Worldnet becomes feasible. Basic Concepts All users are known by three variables: userid, nodename, and network. A userid can be the person s initials, or the person's last name, or anything else the person decided upon when he opened his computer account. A nodename is also known as a hostname. It designates the computer the user is using. The network indicates which of the two dozen or so networks the computer is connected to. If you look at my name at the top of this article, you will see that my userid is Hank, my nodename is Taunivm (that is in Israel, in case you were wondering), and my network is called Bitnet. The nodename and network section of a user's "handle" has been undergoing a transformation in the past few years and this will be explained later. The one common protocol that all networks talk is something called RFC822 standard mail. Within individual networks there are other protocols which will be covered where necessary. ARPANET This network is based on a protocol called TCP/IP. (I know there are people out there reading this and saying, "What does TCP/IP stand for?" But I do not think it is important to know what the letters stand for. When it is important, I will explain it.) It allows for three major applications: FTP, SMTP, and Telnet. FTP stands for File Transfer Protocol and allows a user on one machine to extract a file from any other machine on the network (assuming you know the read password) or allows a user to write a file onto any other machine assuming you know the write password for the destination user and machine. SMTP stands for Simple Mail Transfer Protocol and allows users to send electronic mail almost anywhere in the world. Telnet is a remote-login application. It is not Telenet. But it does basically the same thing. You specify the machine you want to log in to, and Telnet makes the connection from your machine to the one you specified. Most links within ARPANET are 56KB leased lines although there are cases where it may be higher or lower. There are other networks that are modeled after ARPANET: CSnet (Computer Science network), Nsfnet (National Science Foundation Network - which interconnects all supercomputers in the United States), and a few smaller ones. CSnet, up until recently, used primarily X.25 connections via Telenet to establish a connection. They are now switching more and more links over to leased telephone lines. Nsfnet uses primarily T1 lines, which run at 1MB per second. In case you were wondering, ARPANET stands for Advanced Research Projects Agency and is owned by the U.S. government. All of these networks use the TCP/IP protocol and are therefore part of an ever-growing Internet. Bitnet This network spans 27 countries (U.S.A., Canada, West Germany, France, Italy, the Netherlands, Finland, Denmark, Spain, Turkey, Israel, Japan, Mexico, Taiwan, to name a few) and has more than 1,800 computers interconnected. It uses a protocol different than ARPANET but the one common language they talk is electronic mail (RFC822). The European segment of the network is called EARN (European Academic Research Network) and the Canadian section is called NetNorth. All links within Bitnet/EARN/-NetNorth are 9600 baud leased lines. Bitnet stands for Because It's There or Because It's Time. It all depends on who you ask. Bitnet is not the largest network by computer hosts, but is the largest by number of connected countries. If you are an academic institution or a research lab, all you need to do is pay a membership fee per year to Bitnet, Inc. (varies between $1,000 and $10,000) and order a leased line from Telco to your nearest neighbor that has a connection to Bitnet. UUCP Unix to Unix Copy Program Network is a freewheeling, anarchy-type network. It is unknown how many computers are connected to this network but estimates vary from 4,000 to 10,000. Lately, some organizers are trying to put some order into UUCP. It is a slow and grueling process but one that I hope they will succeed at. It has the worst reputation for mail delivery, where delays can be sometimes a week and it is not infrequent that the system loses the mail. Others Here is a brief list of some of the other networks that share RFC822 mail: MFENET: Magnetic Fusion Energy Network; SPAN: Space Physics Analysis Network; JANET: England's National Academic Network; VNET: IBM's corporate internal network; Easynet: DEC's corporate internal network; EUnet: European section of UUCP. There are many other smaller networks that are starting to get off the ground, but as you will see later on, the world of networking is moving away from the concept of an "xxxxNet" to one that imposes a hierarchical structure on all networks. When you add up all the networks and all the machines that can exchange RFC822 mail, the number of machines (from a VAX 730 up to a Cray X/MP) approaches 20,000. Some of the larger systems have 50,000 registered users on their systems while more typically it is around 2,000 users. That means that as a rough estimate, there are about 40 million users that are accessible via RFC822 mail. This grows even larger when you consider that there are experimental gateways that allow networks like Dialcom and MCI Mail to pass RFC822 mail into the Internet and vice versa (no, I will not tell you where they are or how to use them). Most of the users are students, professors, academics, researchers, and school administration personnel. The number of corporate users, like IBM's 200,000 Vnet users, only makes up about 10 percent of the network. What makes this Worldnet system so attractive is that for a large part it is free to use. The university or the company pays Telco for a leased line and connects to the network of their choice. The users of the newly connected computer are then given free access to the network (certain universities impose access restrictions on their users). European sites will soon be undergoing a severe hardship. Their PTTs will require volume charging, so each site will have to restrict usage by their users. At present charging by European PTTs is still on a leased line monthly cost. Since it is a free system, abuse is closely monitored. For example, it is considered bad manners to start a chain letter in the network, since it can quickly grow to saturate the network. Users are caught and in general they understand that disrupting the network will only cause their "free" and genuine mail to be delayed also. Addresses Now for a brief tutorial on how to read network addresses. All RFC822 mail addresses are composed of a LHS and a RHS (Left Hand Side and Right Hand Side). You look at the address and scan for an @-sign. This is the separator between the LHS and the RHS. The LHS is considered the local part of the address. Examples: Hank John Smith steve%hbo.HAIRNET philco!sun!munarri!john These are all samples of LHS addresses. The first two are simple userids. The third one is a gateway. It says that there is an indirect network called HAIRNET that has a machine on it called hbo and you wish to contact the user named steve. The %-sign is used as a kludge to indicate indirect addressing via a gateway that is not directly addressable from all over the Worldnet. The last example is one of UUCP addressing. It reads from left to right. With standard RFC822 addresses, you do not need to know the path the mail will take to get to its final destination. The system takes care of that. UUCP is dumb in that respect. You need to know the path the mail will take. So example 4 says to send it to a machine called philco, which will send it to a machine called sun, which in turn will send it to a machine called munarri, which has a user called john. You can see why people hate UUCP addressing. This type of "bang" addressing is slowly being phased out for the new style of addressing detailed below. But there are still many UUCP sites that prefer their "old" ways. Then again, there are still a lot of people who like Cobol. Here are some examples of a RHS address: taunivm.bitnet wiscvm.wisc.edu relay.cs.net decwrl.dec.com vax.camb.ac.uk vm1.tau.ac.il The first is an example of the old style of addresses - taunivm.bitnet. It is a nodename and a network identifier. The next three are examples of ARPANET addresses. They read from right to left and are tree based. The right-most token represents the higher authority, such as EDU (educational), .NET (network information center), or .COM (commercial). It no longer makes a difference if wiscvm.wisc.edu resides in ARPANET or Bitnet or CSnet. It may indeed be directly connected to all three. The user shouldn't care what network the end user is connected to. Imagine if your friend was connected to Sprint while you used ATT. It shouldn't make a difference in your dialing to know that the end destination is being serviced by Sprint. Just dial the number. That is the concept of "dotted domain names." As soon as you leave the United States, things get even more organized. Every country has an ISO (International Standards Organization) country code. Within each country, an authority decides what second level domain names to assign - such as .AC (academic), .RD (research and development), .COM (commercial), etc. As you move from the right to left of the RHS address, you move from the macro to the micro. Once again, it is important to note that the concept of what network the user resides on becomes a "thing of the past." Putting it all together, we end up with addresses that might look like these: Hank@vm1.tau.ac.il John Smith@decwrl.dec.com steve%hbo.HAIRNET@relay.cs.net In conclusion, the Worldnet supplies electronic mail traffic for free to users with an account on any machine that is connected to one of the networks listed above. The institution ends up picking up the bill for the leased line, while the user only gets charged for the local CPU time and connect time used to create and send the letter. Abuse (chain letters, mass mailings, commercial use of the network, etc.) is frowned upon by the ones who run the networks as well as the hackers who make use of them. If you use the network, don't abuse it. For further reading: Communications of the ACM, October 1986, Notable Computer Networks, Quartermain and Hoskins.