FreeBSD Corporate Networker's Guide, The

Software that is shipped with the source code and contains a software license that disallows mere use of the software in a commercial environment is not Open Source software. The FreeBSD Project does not use such software in FreeBSD because this practice would place most FreeBSD end users in legal trouble.

PC Server and PC Local Area Network

In the old days of IBM XT-compatibles, building a fileserver on a PC was impractical as well as unbelievable. The 8088 hardware simply was not powerful enough, and the XT was extremely restricted in internal bandwidth and other resources. Back then, servers were big, powerful computers that sat in a glass house, if the organization had them at all.

As PCs became more powerful and Ethernet networks extended to all desktops, the Intel 80286 chip began to be used in the AT computer. Because these PCs were constructed with 8MB and 16MB of RAM and 300MB ESDI disks, their use as network servers became possible. These early machines were still very weak compared to a real UNIX server of the time, but because they were much cheaper, network operating systems began appearing--for example, NetWare and LanManager based on OS/2. Companies built large networks based entirely around these NOSs; these networks came to be known as PC-LANs since clients and servers were both PC-compatible computers. The primary difference between a PC used as a server and a PC used as a client was that the server was more powerful, with larger disks, more RAM, and a faster central processing unit (CPU). In addition to this, while PCs were becoming powerful enough to be used as servers, the original proprietary server computers also continued to become more powerful.

Today, vendors selling servers can be selling, in effect, souped-up PCs, or proprietary-hardware computers such as Sun Sparcs. In this book, the term PC server is used to designate a server built around a PC computer (e.g., Wintel) rather than a server computer built with proprietary hardware (e.g., Sparc).

The Token Ring, Latticenet, and Arcnet network media types originally had some marketshare, but Ethernet rapidly became the dominant network media. FreeBSD does support fiber distributed data interface (FDDI) network cards, but the Ethernet network standard is assumed in this book because it is what most 10BaseT and 100BaseT networks are made up of.

Conventions Used

Over the years, the various Microsoft OSs have developed nicknames, although, according to Microsoft, the proper way to refer to them is to use their full names. This book is not an advertisement for MS products, therefore I do not use spelled-out product names for Windows, such as Windows NT Advanced Server, numerous times in a paragraph. Using full names would not only be very tiring for the reader, but it would also make the text read like Microsoft advertising copy. So, here are the shortened terms that are used:

• MS--Microsoft Corporation
• Win31--Microsoft Windows 3.0 and Microsoft Windows 3.1
• WfW--Windows for Workgroups 3.1 and Windows for Workgroups 3.11
• Win16--All Windows 3.0, 3.1, 3.11, Windows for Workgroups 3.1 and 3.11
• Win95, Win98, or Win95/98/ME--Windows 95, Windows 98, Windows Millennium
• NTWKS 3.51, NTWKS 4.0--Windows NT Workstation 3.51, 4.0
• NT Server 3.51, 4.0--Windows NT Server 3.51, 4.0 and Windows NT Advanced Server 3.51, 4.0 (This book doesn't differentiate Advanced from regular NT Server.)
• NT--used when there is no difference in the behavior of the NT Workstation and Server, as well as numeric versions
• Win2K--Windows 2000 Professional (successor to Windows NT Workstation)
• Win2K Server--Windows 2000 Server (successor to Windows NT 4.0 Server)

Normal text in this book is in Times New Roman. Text that is typed into the computer, such as commands, is represented in Courier. Bold Courier indicates computer output. You need to understand that space characters are just as important in command strings as they are in text characters; when typing commands, include the spaces. Special emphasis and keywords are represented in italics. In UNIX, the command interpreter assigns special meanings to double and single quotes. Text that is to be typed into the computer is always exactly what is to be typed, including all forward slashes, backslashes, and/or quote marks.

All URLs in this book are set using underlined Courier (e.g., http://www.freebsd.org). URLs either follow the convention of linking to the root index or linking to the base of the item in question. For example, a Web site may have a separate area for a particular program that is substantively different from the rest of the site, in which case it's apparent that the Webmaster regards the separate area as being apart from the main Web site.

UNIX uses the forward slash (/) as the directory separator in the filesystem, not the backslash (\), as in DOS and Windows. In most cases, I use the fully qualified pathname so that "a file named /foobar is read as a file named "foobar" located in the root directory."

One other formatting convention I use is quad-X. In a command string where input must be substituted, the XXXX mark will appear, as in the following.

passwd root
Changing local password for root.
New password: XXXX
Retype new password: XXXX
passwd: rebuilding the database...
passwd: done

Obtaining RFC Documents

TCP/IP is an open protocol originated by the Internet Engineering Task Force (IETF). This body creates standards through the Request for Comments (RFC) process. In addition to the TCP/IP protocol, RFCs exist for many programs found in UNIX, especially network programs. RFCs are identified by numbers, and this book contains references to specific RFCs. A copy of any RFC can be downloaded from the Internet. Years ago, the master location was ftp://ds.internic.net/rfc. So many books and other publications linked to this machine name, however, that it was shut down. In any case, the Internic is not the network citizen it once was. Today, the master RFC site is http://www.rfc-editor.org/. The most popular primary repository is located at ftp://ftp.isi.edu/in-notes. There is a second repository at http://www.ietf.org/, but its RFC page lists the RFC editor's site as the primary one. It contains mainly text versions along with some PostScript versions. The PostScript versions of the RFCs are better than the text ones because some contain diagrams.

Software and Hardware Mentioned in This Book

In keeping with the spirit of the FreeBSD operating system, practically all software discussed in this book is free, except for the Windows and DOS operating systems. Where specific needs exist on Windows OSs for which no free software package is available, I have used shareware alternatives, always attempting to use the least expensive shareware possible.

Many commercial software packages that manage UNIX-to-Windows interoperability would probably work just as well or better with FreeBSD and Windows desktops. For TCP/IP clients for Windows in particular, the market could use a $20-per-seat Network File System (NFS) client for Win95/98, although this is becoming a moot issue with the release of Win2K.

In some cases, buying hardware devices may be far cheaper and make more sense for many applications than would running software packages on a PC. You owe it to yourself to investigate these as well. Nothing in this book should be construed as a licensing statement for any specific software package. In the case of shareware and freeware, the licenses in the packages take precedence over any statements in this book. It is the end user's responsibility to make sure that all licensing requirements are properly met.

I mention various computer hardware devices that facilitate use of FreeBSD as a network server. Rest assured that I have not received any of these devices free as promotions from their manufacturers, which would influence my selection as to their suitability for use with the operating system. I have purchased all the hardware mentioned here, or used it at businesses that have employe me, where I used the FreeBSD operating system.

Information Sources

An administrator can use a number of significant recognized sources of information about FreeBSD, including the following.

• The manual. All UNIX systems have a system manual, accessed by the man command.
• Electronic documentation. Many document directories can be found in /usr/share, such as the FreeBSD Handbook. If full sources are installed, detailed documentation is often available in source directories, along with the source code.
• The FreeBSD Web site. http://www.freebsd.org
• Mailing lists. The FreeBSD group manages a number of mailing lists, subscription instructions, and a list of mailing lists on the FreeBSD Web site.
• Usenet news. Several Usenet news groups cover FreeBSD.

Usenet is an excellent source of quick information on FreeBSD, but new users should be aware of certain aspects of Usenet and its mailing lists. All users should read the "Netiquette" document in the newsgroup news. newusers before posting. In addition, please keep the following in mind.

• Usenet is not a manufacturer's technical support forum. If you post a question, nobody has any responsibility to respond, or to respond with correct information.
• Usenet is not very fast. In general it may take more than a week before your question is answered by those in the reading group who want to answer it.
• Before posting a question on Usenet, make sure that it is not answered in documentation or on the FreeBSD Web site in the documentation archives. In addition, check past postings at http://www.dejanews.com or some other major Usenet news archive.
• Most readers of the Usenet site visit it irregularly. The one person who can answer your question may not happen to be reading on the few days that you post. Post a question again after a suitable interval, such as a week, if no answer is posted.
• Don't say "send responses by e-mail" in your posting message. Most readers who answer questions on Usenet get annoyed at this because it indicates someone who is too lazy or too busy to bother to follow up on a query. An unreliable news server is not an excuse because DejaNews can be checked for possible missing postings.
• Don't post general questions asking for help. Make your posting as specific as possible. List all relevant details, such as machine, brand, disk size, disk model, amount of memory, exactly what you are trying to do, and so on.
• Always remember that people answering questions on Usenet are doing it for their own enjoyment and advancement of the FreeBSD Project. Your education is of secondary importance.

Megabytes and Megabits

From almost the beginning of the computer industry, computer engineers have measured data by bits. A single bit in a computer can be either on or off, represented by decimal 0 or 1. Bits are grouped together into bytes, which are eight bits long. Bytes are grouped into words, then double words, and so on. All of these units are used with the octal (base 8) or the hexadecimal (base 16) numbering system.

People don't think in hexadecimal, however; they think in the decimal system (base 10). So do most engineers in other disciplines. Only software engineers can think in hexadecimal. So, bits and bytes are often converted from their hexadecimal and octal measurements into decimal by adding the powers of two: two to the zeroeth, plus two to the first, plus two to the second, and so on.

Two raised to the tenth is equal to 1,024. Very early on, software engineers decided to make things easier for themselves and began using the International System of Units (SI) designations--kilo, mega, giga, and so on. So, eight bits is a byte, 1,024 bytes is a kilobyte, 1,024 x 1,024 bits is a megabyte, and so on. Soon, the abbreviation KB came to be associated with kilobytes, MB with megabytes, GB with gigabytes, and so on.

When networking became important to the industry, confusion arose, so people began writing Kbt, Mbt, Gbt--kilobits, megabits, and gigabits--to satisfy the networking and telecommunications people. These people use the decimal measurement, generally, and to them 1Mbt means 1,000,000 bits, not 1,048,576 bits (a megabyte).

Everyone was pretty satisfied with the arrangement and things went along well enough until hard drive manufacturers came along. The problem was that generally the more bits a hard drive holds, the more expensive it is. So, to gain marketshare, hard drive manufacturers began to market their products based on the decimal measurement of bytes the drives could hold. Thus, instead of a 100MB hard drive holding 100 x 1,024 bytes, it actually held only 100 x 1,000 bytes. This allowed hard drive manufacturers to advertise a 10MB drive that didn't actually hold 10MB of data from the computer user's point of view.

As long as hard drives were fairly small, every computer user pretty much accepted the fact that a hard drive's usable capacity was somewhat smaller than its advertised capacity. In the 1990s, however, as hard drives became larger and larger, the advertised capacity began to diverge tremendously from the real capacity. The drive manufacturers, being hardware people, began complaining to the umbrella standards organization--the Institute of Electrical and Electronics Engineers (IEEE)--to do something about it.

Things came to a head, so in December of 1998, the IEEE (which the networking TCP/IP community had rejected in favor of IETF over the IP standards wars) decided to get involved. With a coalition of physicists and measurement purists, they persuaded the International Electrotechnical Commission (IEC) to issue a standard for data storage based on the decimal system. The new standard renamed all of the standard terminology used by the computer industry, with the exception of the hard drive manufacturers, into the base 10 terminology.

In this book, I do not use the Revised Data Processing measurements promoted by the IEC. Because this book is about a computer operating system, the classic measurement system used by software engineers is used except when referring to data transmission--the telecommunications people have never used the computer measurement system. Sadly, the IEC justifies its standard by saying that more people don't know that 1K means 1,024 rather than 1,000. Therefore, the technical people should bow to the majority. An explanation of the IEC's views on this is located at http://physics.nist.gov/cuu/Units/binary.html.

Reaching the Author and Errata

I am always interested in comments and questions from readers. I've set up an e-mail address for comments about the book, and I hope to incorporate readers' suggestions into a future edition. Please refrain from sending general FreeBSD questions until you have read the entire book, asked questions on one of the FreeBSD mailing lists, and searched the archives--http://www.freebsd.org. Unfortunately, I don't have the time (or money) to be a full-time, unpaid, technical support person for FreeBSD. Please send any book comments to book@freebsd-corp-net-guide.com. Naturally, the publisher and I have tried to eliminate all typos and errors from the text. However, a book this size undoubtedly contains at least one mistake. I'll post any necessary corrections you send me on a Web site for this book. Check the Addison-Wesley site and see http://www.freebsd-corp-net-guide.com for postproduction information.

Web Site URLs

No body of case law covers URL linking. I am aware that some people would like to make some sort of global, or at least federal, requirement that anyone linking to a URL first seeks permission from the Web site host. This idea may sound reasonable, but it is not. Because no right to control URL linking exists currently, seeking permission would merely confer a right that never existed to begin with.

In all URLs listed in this book, I have followed any instructions that are posted on the Web site by the Webmaster. However, a Web site that is intentionally made accessible to the public is fundamentally a public venue. If you as a Webmaster do not want others to link to a site that is plainly intended to be public, I respectfully suggest that you examine the definition of public and install IP number block filtering to prevent the general public from viewing the site. I believe that if you place a Web site on the publicly accessible Internet, then people should have the right to link to it, which is exactly the same as their right to visit it. If the Webmaster has placed a directive on the Web site requesting that links to it be made to a particular location, I follow these directives for the URLs in this book.

Because of the fluidity of the Internet, a book that names a significant number of URLs is likely to refer to some sites that have vanished or moved. I apologize for this and I'll try to keep a table of corrected URLs and links available at http://www.freebsd-corp-net-guide.com.