The history of the PC is an unbelievable story, full of successes and failures. Many people who used some of the computer systems before the IBM PC was developed, wipe a tear from their eyes, for various reasons, when they remember their first introduction to computers, typically with the Sinclair Spectrum or the Apple II. In those days, all your programs could be saved to a single floppy disk, 128kB of memory was more than enough to run any program, and the nearest you got to a GUI was at the adhesives shelf at your local DIY store. It must be said that computers were more interesting in those days. Open one up, and it was filled with processor chips, memory chips, sound chips, and so on. You could almost see the thing working (a bit like it was in the days of valves). These days, computers lack any soul; one computer is much like the next. There’s the processor, there’s the memory, that’s a bridge chip, and, oh, there’s the busses, that’s it.

            As we move to computers on a chip, they will, in terms of hardware, become even more boring to look at. But, maybe I’m just biased. Oh, and before the IBM PC, it was people who made things happen in the computer industry, such as William Shockley, Steve Jobs, Kenneth Olson, Sir Clive Sinclair, Bill Gates, and so on. These days it is large teams of software and hardware engineers who move the industry. Well, enough of this negative stuff. The PC is an extremely exciting development, which has changed the form of modern life. Without its flexibility, its compatibility, and, especially, its introduction into the home, we would not have seen the fast growth of the Internet.

            Here is my Top 15 successes in the computer industry:

 

1.      IBM PC (for most), which was a triumph of design and creativity. One of the few computer systems to ever to be released on time, within budget, and within specification. Bill Gates must take some credit in getting IBM to adopt the 8088 processor, rather than 8080. After its success, every man and his dog had a say in what went into it. The rise of the bland IBM PC was a great success of an open-system over closed-systems. Companies who have quasi-monopolies are keen on keeping their systems closed, while companies against other competitors prefer open systems. The market, and thus, the user, prefers open-systems.

2.      TCP/IP, which is the standard protocol used by computers communicating over the Internet. It has been designed to be computer independent to any type of computer, can talk to any other type. It has withstood the growth of the Internet with great success. Its only problem is that we are now running out of IP addresses to grant to all the computers that connect to the Internet. It is thus a victim of its own success.

3.      Electronic mail, which has taken the paperless office one step nearer. Many mourned the death of the letter writing. Before email, TV and the telephone had suppressed the art of letter writing, but with email it is back again, stronger than ever. It is not without its faults, though. Many people have sent emails in anger, or ignorance, and then regretted them later. It is just too quick, and does not allow for a cooling off period. My motto is: ‘If your annoyed about something. Sleep on it, and send the email in the morning’. Also, because email is not a face-to-face communicate, or a voice-to-voice communication, it is easy to take something out of context. So another motto is: ‘Careful read everything that you have written, and make sure there is nothing’. Only on the Internet could email address be accepted, world-wide, in such a short time.

4.      Microsoft, who made sure that they could not loose in the growth of the PC, by teaming up with the main computer manufacturers, such as IBM (for DOS and OS/2), Apple (for Macintosh application software) and for their own operating system: Windows. Luckily for them it was their own operating system which became the industry standard. With the might of having the industry-standard operating system, they captured a large market for industry-standard application programs, such as Word and Excel.

5.      Intel, who was gifted an enormous market with the development of the IBM PC, but have since invested money in enhancing their processors, but still keeping compatibility with their earlier ones. This has caused a great deal of hassle for software developers, but is a dream for users. With processors, the larger the market you have, the more money you can invest in new ones, which leads to a larger market, and so on. Unfortunately, the problem with this is that other processor companies can simply copy their designs, and change them a little so that they are still compatible. This is something that Intel have fought against, and, in most cases have succeed in regaining their market share, either with improved technology or through legal action. The Pentium processor was a great success, as it was technologically superior to many other processors in the market, even the enhanced RISC devices. It has since become faster and faster.

6.      6502 and Z80 processors, the classic 16-bit processors which became a standard part in most of the PCs available before the IBM PC. The 6502 competed against the Motorola 6800, while the Z80 competed directly with the Intel 8080.

7.      Apple II, which brought computing into the class room, the laboratory, and, even, the home.

8.      Ethernet, which has become the standard networking technology. It is not the best networking technology, but has survived because of its upgradeabliity, its ease-of-use, and its cheapness. Ethernet does not cope well with high capacity network traffic. This is because it is based on contention, where nodes must contend with each other to get access to a network segment. If two nodes try to get access at the same time, a collision results, and no data is transmitted. Thus the more traffic there is on the network, the more collisions there are. This reduces the overall network capacity. However, Ethernet had two more trump cards up its sleeve. When faces with network capacity problems, it increased its bit rate from the standard 10Mbps (10BASE) to 100Mbps (100BASE). So there was ten times the capacity which reduced contention problems. For networks backbones it also suffered because it could not transmit data fast enough. So, it played its next card: 1000BASE. This increased the data rate to 1Gbps (1000MBps). Against this type of card player, no other networking technology had a chance.

9.      WWW, which is often confused with the Internet, and is becoming the large, database ever created (okay, 99% of it is rubbish, but even if 1% is good then its all worth while). The WWW is one of the uses of the Internet (others include file transfer, remote login, electronic mail, and so on).

10.  Apple Macintosh, which was one of few PC systems which competed with the IBM PC. It succeeded mainly because of its excellent operating system (MAC OS), which was approximately 10 years ahead of its time. Possibly if Apple had spent as much time in developing application software rather than for their operating system it would have considerably helped the adoption of the Mac. Apple refusing to license it to other manufacturers also held its adoption back. For a long time it thus stayed a closed-system.

11.  Compaq DeskPro 386. Against all the odds, Compaq stole the IBM PC standard from the creators, who had tried to lead the rest of the industry up a dark alley, with MCA.

12.  Sun SPARC, which succeed against of the growth of the IBM PC, because of its excellent technology, its reliable Unix operating system, and its graphical user interface (X-Windows). Sun did not make the mistakes that Apple made, and allowed other companies to license their technology. They also supported open systems in terms of both the hardware and software.

13.  Commodore, who bravely fought on against the IBM PC. They released mainly great computers, such as the Vic range and the Amiga. Commodore was responsible for forcing the price of computers.

14.  Sinclair, who, more than any other company, made computing acceptable to the masses. Okay, most of them had terrible membrane keyboards, and memory adaptor that wobbled, and it took three fingers to get the required command (Shift-2^nd Function-Alt-etc), and it required a cassette recorder to download program, and it would typically crash after you had entered one thousand lines of code. But, all of this aside, in the Sinclair Spectrum they found the right computer, for the right time, at the right price. Sometimes success can breed complacency, and so it turned out with the Sinclair QL and the Sinclair C-5 (the electric slipper).

15.  Compaq, for startling growth, that is unlikely to be ever repeated. From zero to one billion dollars in five years. They achieved they growth, not by luck, but by shear superior technology, and with the idea of sharing their developments.

 

Other contenders include Hewlett-Packard (for their range of printers), CISCO (for their networking products), Java (for ignoring the make of the computer, and its network, and, well, everything), the Power PC (for trying to head off the PC, at the pass), Dell notebooks (because I’ve got one), the Intel 80386, the Intel Pentium, Microsoft Visual Basic (for bring programming to the masses), Microsoft Windows 95, Microsoft Windows NT, and so on. Okay, Windows 95, Windows NT, the 80386 and the Pentium would normally be in the Top 10, but, as Microsoft and Intel are already there, I’ve left them out. Here’s to the Wintel Corporation. We are in their hands. One false move and they will bring their world around themselves. Up to now, Wintel have made all the correct decisions.

 

When it comes to failures, there are no failures really, and it is easy to be wise after the even. Who really knows what would have happened if the industry had taken another route. So instead of the Top 15 failures, I’ve listed the following as the Top 15 under-achievers (please forgive me for adding a few of my own, such as DOS and the Intel 8088):

 

1.      DOS, which became the best selling, standard operating systems for IBM PC systems. Unfortunately, it held the computer industry back for at least ten years. It was text-based, command-oriented, had no graphical user interface, could only access up to 640KB, it could only use 16 bits at a time, and so on, …. Many with a short memory will say that the PC is easy to use, and intuitive, but they are maybe forgetting how it used to be. With Windows 95 (and to a lesser extent with Windows 3.x), Microsoft made computers much easier to use. From then on, users could actually switch their computer on without have to register for a high degree in Computer Engineering. DOS would have been fine, as it was compatible with all its previous parents, but the problem was MAC OS, which really showed everyone how a user interface should operate. Against this competition, it was no contest. So, what was it? Application software. The PC had application software coming out of its ears.

2.      Intel 8088, which became the standard processor, and thus the standard machine code for PC applications. So why is it in the failures list? Well, like DOS, its because it was so difficult to use, and was a compromised system. While Amiga and Apple programmers were writing proper programs which used the processor to its maximum extent, PC programs were still using their processor in ‘sleepy-mode’ (8088-compatiable mode), and could only access a maximum of 1MB of memory (because of the 20-bit address bus limit for 8088 code). The big problem with the 8088 was that it kept compatibility with its father: the 8080. For this Intel decided to use a segmented memory access, which is fine for small programs, but a nightmare for large programs (basically anything over 64KB).

3.      Alpha processor, which was DEC’s attack on the processor market. It had blistering performance, which blew every other processor out of the water (and still does). It has never been properly exploited, as there is a lack of development tools for it. The Intel Pentium proved that it was a great all-comer and did many things well, and was willing to improved the bits that it was not so good at.

4.      Z8000 processor, which was a classic case of being technically superior, but was not compatible with its father, the mighty Z80, and its kissing cousin, the 8080. Few companies have given away such an advantage with a single product. Where are Zilog now? Head buried in the sand, probably.

5.      DEC, who were one of the most innovate companies in the computer industry. They developed a completely new market niche with their minicomputers, but they refused, until it was too late, that the microcomputer would have an impact on the computer market. DEC went from a company that made a profit of $1.31 billion in 1988, to a company which, in one quarter of 1992, lost $2 billion. Their founder, Ken Olsen, eventually left the company in 1992, and his successor brought sweeping changes. Eventually, though, in 1998 it was one of the new PC companies, Compaq, who would buy DEC. For Compaq, DEC seemed a good match, as DEC had never really created much of a market for PCs, and had concentrated on high-end products, such as Alpha-based workstations and network servers.

6.      Fairchild Semiconductor. Few companies have ever generated so many ideas and incubated so many innovative companies, and got little in return.

7.      Xerox. Many of the ideas in modern computing, such as GUIs and networking, were initiated at Xerox’s research facility. Unfortunately, Xerox lacked force to develop them into products, maybe because they reduced Xerox’s main market, which was, and still is, very much based on paper.

8.      PCjr, which was another case of incompatibility. IBM lost a whole year in releasing the PCjr, and lost a lot of credibility with their suppliers (many of whom were left with unsold systems) and their competitors (who were given a whole year to catch-up with IBM).

9.      OS/2, IBM’s attempt to regain the operating system market from Microsoft. It was a compromised operating system, and their development team lacked the freedom of the original IBM PC development. Too many people and too many committees were involved in its development. It thus lacked the freedom, and independence that the Boca Raton development team had. IBM’s mainframe divisions were, at the time, a powerful force in IBM, and could easily stall, or veto a product if it had an effect on their profitable market.

10.  CP/M, which many believed would because the standard operating system for microcomputers. Digital Research had an excellent opportunity to make it the standard operating system for the PC, but Microsoft overcame them by making their DOS system much cheaper.

11.  MCA, which was the architecture that IBM tried to move the market with. It failed because Compaq, and several others, went against it, and kept developing the existing architecture.

12.  RISC processors, which were seen as the answer to increased computing power. As Intel as shown, one of the best ways to increase computing speed is to simply ramp-up the clock speed, and make the busses faster.

13.  Sinclair Research, who after the success of the ZX81 and the Spectrum, threw it all away by releasing a whole range of under-achievers, such as the QL, and the C-5.

14.  MSX, which was meant to be the technology that would standardize computer software on PCs. Unfortunately, it hadn’t heard of the new 16-bit processors, and most of all, the IBM PC.

15.  Lotus Development, who totally misjudged the market, by not initially developing their Lotus 1-2-3 spreadsheet for Microsoft Windows. They instead developed it for OS/2, and eventually lost the market leadership to Microsoft Excel. Lotus also missed an excellent opportunity to purchase a large part of Microsoft when they were still a small company. The profits on that purchase would have been gigantic.

 

So was/is the IBM PC a success? Of course it was/is. But, for IBM it has been a double-edged sword. It opened-up a new, and exciting market, and made the company operate in ways that would have never been possible before. Before the IBM PC, their systems sold by themselves, because they were made by IBM. It also considerably reduced their market share. Many questions remained unanswered: ‘Would it have been accepted in the same way if it had been a closed system, which had to be licensed from IBM’, ‘Would it have been accepted if it had used IBM components rather than other standard components, especially the Intel processors’, ‘Would they have succeeded in the operating system market if they had written DOS by themselves’, and so on.  Who knows? But, from now on we will refer to the computers based on the x86 architecture as PC’s.

            Oh, and as an academic I would like to give a special mention to the C programming, which has given me great heartaches over the years. Oh, yeah, it’s used extensively in industry and is extremely useful. It is the programming language that I would automatically use for consultancy work. C is well supported by the major language package developers, and there are a great deal of code available for it. But for teaching programming, it is a complete non-starter.  Without going into too much detail, the problems with C are not to do with the basic syntax of the language. It’s to do with a thing called pointers. They are the most horrible things imaginable when it comes to teaching programming languages, and they basically ‘point’ to a location in memory. This is fine, but in most cases you don’t really have to bother about where in memory things are stored. But, C forces you to use them, rather than hiding them away. So, in a C programming class, things go very well until about the 8^th week when pointers are introduced, and then that’s it. Oh, and don’t get me started on C++.

So, what is it that differentiates one PC system from another? What makes one better than another? It is difficult to say, but basically its all about how well bolted together systems are, how compatible the parts are with the loaded software, how they organise the peripherals, and so on. The big problem though is compatibility, and compatibility is all about peripherals looking the same, that is, having the same IRQ, the same I/O address, and so on.

            The PC is an amazing device, and has allowed computers to move from technical specialists to, well, anyone. However, they are also one of the most annoying of pieces of technology of all time, in terms of their software, their operating system, and their hardware. If we bought a car and it failed at least a few times every day, we would take it back and demand another one. When that failed, we would demand our money back. Or, sorry I could go on forever here, imagine a toaster that failed half way through making a piece of toast, and we had to turn the power off, and restart it. We just wouldn’t allow it.

            So why does the PC lead such a privileged life. Well it’s because it’s so useful and multitalented, although it doesn’t really excel at much. Contrast a simple games computer against the PC and you find many lessons in how to make a computer easy-to-use, and to configure. One of the main reasons for many of its problems is the compatibility with previous systems both in terms of hardware compatibility and software compatibility (and dodgy software, of course). The big change on the PC was the introduction of proper 32-bit software, Windows 95/NT.

            In the future systems will be configured by the operating system, and not by the user. How many people understand what an IRQ is, what I/O addresses are, and so on. Maybe if the PC faced some proper competition it would become an easy-to-use and become totally reliable. Then when they were switched on they would configure themselves automatically, and you could connect any device you wanted and it would understand how to configure (we’re nearly there, but it’s still not perfect). Then we would have a tool which could be used to improve creativity and you didn’t need a degree in computer engineering to use one (in your dreams!). But, anyway, it’s keeping a lot of technical people in a job, so, don’t tell anyone our little secret. The Apple Macintosh was a classic example of a well-designed computer that was designed as a single unit. When initially released it started up with messages like “I’m glad to be out of that bag” and “Hello, I am Macintosh. Never trust a computer you cannot lift.” 

            So, apart from the IBM PC, what are the what are the all-time best computers. A list by Byte in September 1995 stated the following:

 

1.          MITS Altair8800     2.          Apple II 3.          Commodore PET   4.          Radio Shack TRS-80               5.          Osborne 1 Portable               6.          Xerox Star 7.          IBM PC 8.          Compaq Portable   9.          Radio Shack TRS-80 Model 100 10.     Apple Macintosh

11.     IBM AT 12.     Commodore Amiga 1000          13.     Compaq Deskpro 386               14.     Apple Macintosh II 15.     Next Nextstation      16.     NEC UltraLite           17.     Sun SparcStation 1     18.     IBM RS/6000 19.     Apple Power Macintosh 20.     IBM ThinkPad 701C

 

And the Top 10 computer people as:

 

1.                    DAN BRICKLIN (VisiCalc) 2.                    BILL GATES (Microsoft) 3.                    STEVE JOBS (Apple) 4.                    ROBERT NOYCE (Intel) 5.                    DENNIS RITCHIE (C Programming) 6.                    MARC ANDREESSEN (Netscape Communications) 7.                    BILL ATKINSON  (Apple Mac GUI) 8.                    TIM BERNERS-LEE (CERN) 9.                  DOUG ENGELBART (Mouse/Windows/etc) 10.              GRACE MURRAY HOPPER (COBOL)

11.           PHILIPPE KAHN (Turbo Pascal) 12.           MITCH KAPOR (Lotus 123) 13.           DONALD KNUTH (TEX) 14.           THOMAS KURTZ 15.           DREW MAJOR (NetWare) 16.           ROBERT METCALFE (Ethernet) 17.  BJARNE STROUSTRUP (C++) 18.           JOHN WARNOCK (Adobe) 19.           NIKLAUS WIRTH (Pascal) 20            STEVE WOZNIAK (Apple)

 

One of the classic comments of all time was by Ken Olson at DEC, who stated, “There is no reason anyone would want a computer in their home.” This seems farcical now, but at the time, in the 1970s, there were no CD-ROMs, no microwave ovens, no automated cash dispensers, and no Internet. Few people predicted them, so, predicting the PC was also difficult. But the two best comments were:

 

“Computers in the future may weigh no more than 1.5 tons.” Popular Mechanics

“I think there is a world market for maybe five computers”, Thomas Watson, chairman of IBM, 1943

 

Oh boy, is it confusing. There are so many different busses used in systems, especially in PCs. There are three main reasons for the number of busses: legacy, compatibility and efficiency. With legacy, busses exist because they have existed in the past, and are required to be supported for the time being. With compatibility, busses allow the segmentation of a system and provide, most of the time, a standard interface that allows an orderly flow of data and allows manufactures to interconnect their equipment. If allowed, many manufacturers would force users to use their own versions of busses, but the trend these days is to use internationally defined busses. Efficient busses have been designed to take into account the type of data that is being transmitted. So, in the end, we win.

            Sometimes, though the bus technology does always win, and manufacturers who try to develop systems on their own can often fail miserably. This has been shown with the MCA bus, which was an excellent step in bus technology, and made-up for the mistakes made in the original PC design. But, IBM tried to force the market, and failed. In these days, it is International Standard that are important. Products to be accepted in the market, or in the industry require an international standards body to standardize it, such as the IEEE, the ISO, ANSI, and so on. Without them very few companies will accept the product. A classic case of an open standard winning against a closed system was Betamax video which was a closed standard produced by Sony, was up against VHS which was an open-standard. VHS was the inferior technology, but won because there were more companies willing to adopt it.

 

There’s an amusing statement that was made in 1981, in the book 30 Hour BASIC Standard, 1981:

 

“Microcomputers are the tool of the 80’s. BASIC is the language that all of them use. So the sooner you learn BASIC, the sooner you will understand the microcomputer revolution”

 

Well, as it has proven, a good knowledge of BASIC will not really help your understanding of microcomputers, but, if there is one bus that you need to understand in the PC it is the PCI bus. This is because it is basically the main interface bus within the PC. Most external devices eventually connect to the PCI through bridge devices. Their were several short-term fixes for local bus technology, but the PCI was the first attempt at a properly designed system bus. It allows the PC to be segmented into differing data transfer rates. PCI provides a buffer between the main system core, such as the processor and its memory, and the slower peripherals, such as the hard-disk, serial ports, and so on.

            With interrupts, the PCI has the great advantage over ISA in that it allows interrupts to be properly assigned at system startup. The BIOS or the operating system can communicate with the PCI-connected bus with the configuration address area. From this, the system can determine the type of device it is, whether it be a graphics card or a network card. The system can then properly configure the device and grant it the required resources. The days of users having a assign interrupts (using card jumpers, in some cases) and I/O addresses are reducing (thankfully!).

            There is great potential in the PCI bus. At present, most systems use 32-bit data transfers, but there is scope for 64-bit data transfers. Also, the 33MHz clock can be increased to 66MHz with double edge clocking. A new enhanced PCI-based system called the AGP (Advanced Graphics Port) has been developed which allows for data transfers of over 500MB/s.

            I’m slightly annoyed with the success of the PCI bus, as I’ve got an ISA-based sound card, an ISA-based Ethernet card and an ISA-based video camera, and I’ve only got two ISA slots. So, every so often, I have to swap the sound card for the video camera, and vice-versa. At present, I’ve got four empty PCI slots, and I think one of them is waiting for a PCI-based Ethernet card. Then I’ll be able to have a proper video conference, with sound and video. But, never mind, I’ve just got myself a lovely new Dell notebook, and a USB-based video camera, and a single PCMCIA card for my modem and network connections, so I may never need my desktop computer again (here’s hoping).

The IDE bus. What can you say about it? Not much really. It has no future plans for glory and is looking forward to a graceful retirement. It works, it’s reliable, it’s standard, it’s cheap, blah, blah, and relatively easy to set up. I’ve spent many a happy hour (not!) setting the jumpers on CD-ROM drives and secondary hard-disk drives which I want to add to a PC system. Luckily, these days, modern disk drives and BIOS cope well with adding and deleting disk drives to systems.

            On its negative side, IDE is not really that fast, but it really doesn’t have to be, as disk drives do not require high data rates. E-IDE improved IDE a great deal and only required a simple change in the BIOS. In conclusion, SCSI is the natural choice for disk drives and allows for much greater flexibility in configuration and also high data rates. But, it tends to be more expensive, and we’d miss IDE, wouldn’t we?

SCSI’s full grown-up name is the Small Computer Systems Interface. It is difficult to define exactly what a small computer system is[1], but SCSI has outgrown its original application of interfacing to ‘small’ systems and to external disk drives. It now has the potential of being able to interface virtually any external peripheral to a system. It can also be used to connect devices internally within a system. Typically, it takes a bit longer to initially boot the system, but once it has, it should be as reliable as any non-SCSI device.

            An important concept in SCSI is the prioritization of devices using SCSI IDs. Few busses allow the system to prioritize peripherals. Thus, in a properly configured system, fast devices which require to be quickly serviced will always get access onto the bus before slow devices which do not require fast servicing. Unfortunately, the method SCSI uses limits the number of devices to one less than the number of bits on the data bus (7 for an 8-bit data bus and 15 for a 16-bit data bus). In most cases, this is not a major problem. For example, two hard disks, two CD-ROM drives, a tape backup system, a zip drive and a midi keyboard could all be attached to a standard SCSI-I bus.

            In most PCs the IDE drive is still used in the majority of systems, as it is relatively easy to setup and it’s cheap. It is also dedicated to interfacing to the disk drives; thus, no other peripheral can hog the disk drive bus. However, for most general-purpose applications, SCSI is best. New standards for SCSI give a 16-bit data bus, at a transfer rate of 20MHz, giving a maximum data throughput of 40MB/s, which is much faster than IDE. It is also much easier to configure a SCSI system than it is connecting peripherals internally in a PC. A SCSI system only requires a single interrupt line, for all the devices that are connected.

PCMCIA devices. To save paper, I’ve got seven lines to tell you about them. Well, in summary, they’re really good, but tend to be relatively expensive. The big usage of them is to add a network adapter or a modem to a notebook computer. They are typically not used to add to the memory of the notebook or to increase its hard disk space (an internal upgrade is much better for these). Personally, I find them a little too thin, and I can’t believe they can get all the required electronics into them (but I remember when simple logic ICs, like AND and OR gates, were as big as your thumb and they could heat it if you required).

Congratulations go to the USB port. It was the first truly generic, easy-to-use, connection bus for the PC that has mechanisms for non-real-time (such as printer data) and real-time data (such as video, audio and speech). It allows for the easy addition and removal of devices from the system, and it also supports hot plugging (adding or removing a device while the computer is on). Microsoft first supported USB in Windows 95 OSR2, and it has since become one of the most used ports, for devices such as video cameras, CD-ROM drives, printers, digital speakers, monitors, and so on. The only problem with USB is that it only gives a data throughput of 12Mbps, and thus cannot be used for high-speed devices. Possibly, over time, this rate may be increased, or other faster busses, such as Firewire could be used for high-speed applications, such as Fast Ethernet, multimedia communications, hard disk interfaces, and so on.

The joypad, keyboard and mouse port are all based on a legacy type system. Over time, the USB port should replace each interface type, but as they work well at the present they may be around for a while longer.

            The method that the games port uses to determine position is rather cumbersome, where it uses a single-shot monostable timer to determine the x- and y-positions. An improved method is to pass the data using a serial interface, just as the mouse does. But, it’s a standard, and that’s the most important thing.

            The keyboard and PS/2-style mouse connections have proved popular, as they are both now small 5-pin DIN-style connectors, and as the software automatically scans the port for devices, they can be plugged into either socket. This allows for an extra keyboard or a second mouse to be used with a notebook.

            As I’ve got a few extra lines at the end of this chapter, I would like to review the material that has been covered up to this point. The key to understanding internal busses is contained in the Motherboard chapter, where the processor interfaces with the TXC device, which directs any requests to the second level cache, the DRAM memory or the PCI bus. The PCI bridge device is also important as it isolates the other busses, such as ISA/IDE, USB, serial/parallel port from the PCI bus, and thus the rest of the system. The keyboard, games port and mouse interfaces are accessed via the PCI bridge.

So, what’s the biggest weakness of the PC. In the past, it has probably been the graphics facilities. This is mainly because the bus systems within the PC did not support large data throughput (ISA/EISA is way too slow). The design of the graphics system also required that the video card required to store all the data which was to be displayed on the screen. Thus no matter the amount of memory on the system, it was still limited by the amount of memory on the graphics card. AGP overcomes this by allowing graphical images to be stored in the main memory and then transferred to the video displayed over a fast bus.

            The data demand for graphical displays is almost unlimited, as the faster they can be driven, the greater their application. The AGP bus is an excellent extension to the PCI bus, and gives data throughput of over 500MB/s, whereas standard PCI devices can typically only be run at less than 100MB/s. AGP is now a standard part of most PC motherboards, and it is still to be seen if many systems will start to use this port.

 

Good old RS-232. My Bank Manager would certainly agree with this, as I have made more consultancy income with it than any other piece of computer equipment. I have also ran more RS-232 training courses than all the trendy subjects areas (such as Java and C++) put together (well, anyway, it doesn’t take much to run a C++ course!). The reason for this is because it is one of the least understood connections on computer equipment. I’ve interfaced PCs to gas chromatographs (using an 8-port RS-232 card, heavy!), a PC to a VAX, a Sun workstation to a PC, a PC to another PC, a Honeywell TDC to a PC, a PC to a PLC, and so on. For most applications, a serial port to serial port connection is still the easiest method to transfer data from one computer to another.

            RS-232 is one of the most widely used ‘standards’ in the world. It is virtually standard on every computer and, while it is relatively slow, it is a standard device. This over-rules its slowness, its non-standardness, its lack of powerful error checking, its lack of address facilities, and, well, need I go on. It shares its gold stars with solid performers, such as Ethernet and the Parallel Port. Neither of these are star performers and are far from perfect, but they are good, old robust performers who will outlast many of their more modern contenders. When their position is challenged by a young contender, the standards agency simply invest a lot of experience and brainpower to increase their performance. Who would believe that the data rate, over copper wires, could be increased to 1Gbps for Ethernet to 1MBps for RS-422. One trusted piece of equipment I could have every done without is an RS-232 transmitter/receiver. For this, I used an old 80386-based laptop computer (which weights as much as a modern desktop computers) which ran either a simple DOS-based transmitter/receiver program (see previous chapter), or the excellent Windows 3.1 Terminal program. These I could use just as an electronic engineer would use a multimeter to test the voltages and currents in a circuit. A telltale sign that I was transmitting at the wrong bit rate or using an incorrect number of data bits was the incorrectly received characters (but at least it was receiving characters, which was an important test).

 

The RS-422/RS-485 standard really does enhance the basic RS-232 standard and allows for standard RS-232 connections to be increased to over 1.2km (although at low bit rates and un-noisy environments allows for even greater distances). A surprising thought in these days of cheap electronics, and PC motherboards that cost less than $100, is that RS-232 to RS-485 convertors are relatively expensive. In fact, it is possible to get a complete PC motherboard for the same price as an RS-232/RS485 convertor (which contains just a few internal components), but as the convertor saves lots of time and energy they are probably worth the high costs.

 

 

What a strange device a modem is. It has such as thankless task; converting information from lovely, pure digital signals into something that passes over a speech-limited voice channel. It does its job well and with compression can achieve reasonable results. But, really, it’s a short-term fix, before we all use high-speed connections with proper data cables, whether they be shield twisted-pair cables or fiber optic cables. So, modems allow us to migrate our telephone connection to a full-blown network connection. The motivation for the increased bandwidth is the Internet and especially the integration of fully digital multimedia connections.

            The AT command code allows for a standardization in the modem operation, but as many have seen, modems are not quite as compatible as they may seem.

The parallel port is hardly the greatest piece of technology. In its truly standard form, it only allows for simplex communications, from the PC outwards. However, like the RS-232 port, it’s a standard part of the PC, and it’s cheap. So, interface designers have worked under difficult circumstances to try and improving its specification, such as increasing its bit rate and allowing multiple devices to connect to it at the same time, but it still suffers from a lack of controllability. Anyone who has changed the interface of a device from the parallel port to the USB will know how much better the USB port is over the parallel port.

     The parallel port and RS-232 are the two top requests that I get from users, especially related to project work. The Top 10 requests, in order of the most requests I have received, are:

 

1.     RS-232. 2.     Parallel Port. 3.     Converting a DOS program to Microsoft Windows. 4.     Borland Delphi interfacing. 5.     ISA card design.

6.   Interrupt-driven software. 7.   PCMCIA. 8.   Network card design. 9.   Visual Basic interfacing 10. Using buffered systems.

 

One of the most amusing emails that I ever received related to an ISA card which I had drawn. In the card, I had drawn a few chips, to basically show that it had some electronics on it. So that the chips would not be confused with real chips I labeled one of them XYZ123. One user sent me an email saying:

 

‘Thanks for … Please could you tell me the function of the XYZ123 device. I have searched for this component, and can not find any information on it. Please could you send me some’

 

I didn’t really have the heart to write back to the user and say that it was a made-up chip, so I sent an email back saying that it was not available at the present time (which was true).

 

The parallel port was never really been destined for glory. It is basically a legacy port, which, in the past, was only really useful in connecting printers. The future for printer connections is either with network connections, such as Ethernet, or with a USB connection. In its standard form, it has a large, bulky connector, which in many systems is never even used.

            It has always struggled against the serial port, because it lacks the flexibility of RS-232 and, until recently, had no standards agency to support it. However, it’s there and it has great potential for fast data transfers. RS-232 has always been a great success and has many of the large manufacturers supporting it, and all importantly, it is defined by several standards agencies. The key to its current success was due to the intervention of the NPA which brought together many of the leading computer and printer manufacturers. In these days, there are only a few major companies, such as Intel and Microsoft, who can lead the market and define new standards (such as the PCI bus, with Intel).

            The main difficulties are how to keep compatibility with previous implementations and software, and also how to connect multiple devices on a bus system, and allow them to pass data back and forward without interfering with other devices. This has finally been achieved with ECP/EPP mode. It is a bit complex, but it works, and even supports data compression. At the present, my notebook connects to a CD-R drive, a scanner and a printer, all of the same parallel port (just like SCSI). This arrangement works well most of the time and is a relative cheap way of connecting devices, but it is in no way as flexible and as fast a SCSI.

Modbus is an important protocol and has grown in its popularity because of its simplicity. It has a very basic structure is easy extremely easy to implement as it is based on a master/slave relationship where a master device sends commands and the addressed slave responses back with the required information. Its main advantages are its simplicity, its standardization and its robustness.

            Modbus can be operated on a wide range of computers running any type of software, from a simple Terminal-type connection, where the user can enter the required commands and views the responses, or it can be implemented through a graphical user interface, with the commands and response messages hidden from the user. The basic protocol is of-course limited in its basic specification, such as the limited number of nodes (256, maximum) and the limited addressing range (0000h to FFFFh).

            The basic communications link is also simple to implement (normally, RS-232), but newer Modbus implementations use network connections, such as Ethernet. Another change is to implement the Modbus protocol over a standard TCP/IP-based network. This will allow Modbus to be used over an Internet connection.

            RS-232 does not have strong error checking, and only provides for basic parity check. Modbus using ASCII-based transmission of the Modbus protocol adds a simple checksum to provide an improved error detection technique (LRC). For more powerful error detection the data can be transmitted in RTU format, which uses the more powerful technique (CRC).

            The Modbus Plus protocol now allows for devices to be either a master or a slave. This allows for distributed instrumentation, where any device can request data from any other device, at a given time.

The instrumentation industry has moved over the years from instrumentation networks made from dumb instruments which reported back to a central controller, to smart instruments with distributed control. Fieldbus is really the first true implementer of totally distributed systems, but as the scope of the Fieldbus is limited to areas, there is still a need for a global control system (such as a DCS). The Fieldbus is excellent at allowing local control and parameter conversion, but is not so good at providing a global control strategy. This disadvantage is outweighed by reduced cabling costs, as Fieldbus connects onto a bus, and devices easily connect to the bus.

            Future instrumentation networks have no need to involve a complex main controller, as the instruments themselves have the power to locally control. The function of the main controller will change from getting involved with the low-level operations of instrumentation to the high-level functions of advanced control, interarea control, centralized configuration, alarm filtering and maintaining a global database.

            Serial communication, such as RS-485, has allowed for multidrop serial communication networks, and has proven to be an excellent method of providing a highly-reliable, fast communications channel. But, unfortunately it is still basically a communication channel and most of the higher-level protocols are vendor specific. The Fieldbus is an attempt to overcome this and to provide a standard method, which is well matched to control and data acquisition.

            The days of manufacturers creating a virtual monopoly with vendor-specific is now, thankfully, receding. At one time organizations were generally tied by the vendor of the main control system, this was the only way that they could guarantee compatibility. International standards overcome this problem but forcing manufacturers to conform to the standard. Any vendor who does not conform will quickly loose market share, unless they are a true market leader, and have the power to force the whole industry in a certain direction. Today even the market leaders, such as Honeywell, have to conform to standards and become involved with other companies to develop industry standard, which are then developed as international standards by the relevant standards agency.

 

WorldFIP is an excellent example of a well-designed bus that is simple to setup and use. It uses many of the techniques developed in computer networks, such as the use of Manchester coding and collision detection. It is also based on a layer approach, such as having a physical layer, a data link layer, a management layer, and so on. This fits in well with the OSI 7-layered model that is used in computer networks (see Chapter 25), and allows manufacturers of differ­ent systems to interconnect their equipment through standard interfaces. It also allows software and hardware to integrate well and be portable on differing systems.

            The layered approach also allows for different implementations on each layer. In its current form it supports bitrates of 31.5kbps, 1Mbps, 2.5Mbps and 5Mbps, over copper and fiber optic cables. The polling of data on a WorldFIP network is also extremely flexible where messages can either be sent periodically or aperiodically.

            Another great advantage of WorldFIP is that each parameter on the network can be assigned a unique ID (a tag). As it is a 16-bit field, up to 65,636 of these tags can be used. The addressing of the devices is also powerful, and over 1 million addressable devices is possible (24-bit address).

 

As with the WorldFIP bus, the CAN bus is a well-designed network, based on techniques learned from computer networks. It is a serially connected bus, where all nodes have access to the network, and collisions between nodes are detected within a very short time. This allows devices to have a relatively equal share of the bandwidth of the bus. As automobiles are noisy environments, the CAN bus is a rugged bus which copes well with errors, and also devices which are not operating correctly.

            The relatively high bit rates of the CAN bus allows a great deal of information to be passed between instruments and their controllers. To prevent major problems, the bus can be organized into segments, so that a major fault in one area does not greatly affect other areas. A failure of any of the controllers can lead to major problems, so secondary controllers can be made to monitor the operation of the primary, and can remove the primary controller from the bus if they are not operating correctly. Another method is to allow localized control when the primary control is not functioning properly.

            Power dissipation is also a major factor in cars as devices must be ready to respond quickly to events, but not to dissipate much power when they are idle. Thus, the CAN bus has methods to allow devices to sleep if they are inactive, and then be awoken when a specific event occurs.

            The car of the future, based on the CAN bus, would have little need for complex wiring harnesses, and would simply require the daisy chaining of devices onto the common bus. The connector used can be matched to the environment, such as heavy-duty connector for robust situations, or a light connector for ease of connection/disconnection.

            As the CAN bus has been designed with a thought for the 7-layered OSI model, which is used in computer networks, there is great potential for using standard computer network protocols, such as TCP/IP. Thus will allow CAN busses to connect straight into the Internet, and allow for remote control and remote data acquisition over the Internet, or over a local or wide area network. The data could be protected using data encryption techniques. So, maybe one day you could log into the Internet and switch on the air conditioning in your car before you even leave your house.

The IEEE-488 is a beautifully designed bus, which is well supported by software vendors, and is easy to set-up. It will basically run quietly for many years without requiring any intervention by the user. The connector and cable are very well designed and can stand a great deal of abuse. It has typically been used a standard interface for instrumentation, but the growth of the serial busses is likely to reduce its importance.

            And what can I say about the VME bus. Oh boy, it’s complex. Its little brother, the VXI is a little less complex, but still is an extremely powerful and flexible bus for building modular instrumentation systems. Unfortunately it suffers from being too flexible and can be complex to write software for. The popularity of the PCI bus, and especially the CompactPCI bus (the PCI bus for modular systems) is overtaking the VXI bus.

Which two people have done more for world unity than anyone else? Well, Prof. TCP and Dr. IP must be somewhere in the Top 10. They have done more to unify the world than all the diplomats in the world have. They do not respect national borders, time zones, cultures, industrial conglomerates or anything like that. They allow the sharing of information around the world, and are totally open for anyone to use. Top marks to Prof. TCP and Dr. IP, the true champions of freedom and democracy.

            Many of the great inventions/developments of our time were things that were not really predicted, such as CD-ROMs, RADAR, silicon transistors, fiber optic cables, and, of course, the Internet. The Internet itself is basically an infrastructure of interconnected networks which run a common protocol. The nightmare of interfacing the many computer systems around the world was solved because of two simple protocols: TCP and IP. Without them the Internet would not have evolved so quickly and possibly would not have occurred at all. TCP and IP are excellent protocols as they are simple and can be run over any type of network, on any type of computer system.

            The Internet is often confused with the World Wide Web (WWW), but the WWW is only one application of the Internet. Others include electronic mail (the No.1 application), file transfer, remote login, and so on.

            The amount of information transmitted over networks increases by a large factor every year. This is due to local area networks, wide area networks, of course, traffic over the Internet. It is currently estimated that traffic on the Internet doubles every 100 days and that three people join the Internet every second. This means an eight-fold increase in traffic over a whole year. It is hard to imagine such growth in any other technological area. Imagine if cars were eight times faster each year, or could carry eight times the number of passengers each year (and of course roads and driveways would have to be eight times larger each year).

In this chapter I’ve presented the two opposite ends of code development for TCP/IP communications. The C++ code is complex, but very powerful, and allows for a great deal of flexibility. On the other hand, the Visual Basic code is simple to implement, but is difficult to implement non-typical applications. Thus, the code used tends to reflect the type of application. In many cases Visual Basic gives an easy-to-implement package, with the required functionality. I’ve seen many a student wilt at the prospect of implementing a Microsoft Windows program in C++. ‘Where do I start’, is always the first comment, and then ‘How do I do text input’, and so on. Visual Basic, on the other hand, has matured into an excellent development system which hides much of the complexity of Microsoft Windows away from the developer. So, don’t worry about computer language snobbery. Pick the best language to implement the specification.

            UDP transmission can be likened to sending electronic mail. In most electronic mail packages the user can request that a receipt is sent back to the originator when the electronic mail has been opened. This is equivalent to TCP, where data is acknowledged after a certain amount of data has been sent. If the user does not receive a receipt for their electronic mail then they will send another one, until it is receipted or until there is a reply. UDP is equivalent to a user sending an electronic mail without asking for a receipt, thus the originator has no idea if the data has been received, or not.

            TCP/IP is an excellent method for networked communications, as IP provides the routing of the data, and TCP allows acknowledgements for the data. Thus, the data can always be guaranteed to be correct. Unfortunately there is an overhead in the connection of the TCP socket, where the two communicating stations must exchange parameters before the connection is made, then the must maintain contain and acknowledge received TCP packets. UDP has the advantage that it is connectionless. So there is no need for a connection to be made, and data is simply thrown in the network, without the requirement for acknowledgments. Thus UDP packets are much less reliable in their operation, and a sending station cannot guarantee that the data is going to be received. UDP is thus useful for remote data acquisition where data can be simply transmitted without it being requested or without a TCP/IP connection being made.

            The concept of ports and sockets is important in TCP/IP. Servers wait and listen on a given port number. They only read packets which have the correct port number. For example, a WWW server listens for data on port 80, and an FTP server listens for port 21. Thus a properly setup communication network requires a knowledge of the ports which are accessed. An excellent method for virus writers and hackers to get into a network is to install a program which responds to a given port which the hacker uses to connect to. Once into the system they can do a great deal of damage. Programming languages such as Java have built-in security to reduce this problem.

Many of the great inventions/developments of our time were things that were not really predicted, such as CD-ROMs, RADAR, silicon transistors, fiber optic cables, and, of course, the Internet. The Internet itself is basically an infrastructure of interconnected networks which run a common protocol. The nightmare of interfacing the many computer systems around the world was solved because of two simple protocols: TCP and IP. Without them the Internet would not have evolved so quickly and possibly would not have occurred at all. TCP and IP are excellent protocols as they are simple and can be run over any type of network, on any type of computer system.

            The Internet is often confused with the World Wide Web (WWW), but the WWW is only one application of the Internet. Others include electronic mail (the No.1 application), file transfer, remote login, and so on.

            The amount of information transmitted over networks increases by a large factor every year. This is due to local area networks, wide area networks, of course, traffic over the Internet. It is currently estimated that traffic on the Internet doubles every 100 days and that three people join the Internet every second. This means an eight-fold increase in traffic over a whole year. It is hard to imagine such growth in any other technological area. Imagine if cars were eight times faster each year, or could carry eight times the number of passengers each year (and of course roads and driveways would have to be eight times larger each year).

            Networks have grown vastly over the past twenty years, and most companies now have some form of network. At the beginning of the 1980s, PCs were relatively complex machines to use, and required application programs to be installed locally to their disk drives. Many modern computers now run their application programs over a network, which makes the administration of the application software must simpler, and also allows users to share their resources.

            The topology of a network is all-important, as it can severely effect the performance of the network, and can also be used to find network faults. I have run a network for many years and know the problems that can occur if a network grows without any long-term strategy. Many users (especially managers) perceive that a network can be expanded to an infinite degree. Many also think that new users can simply be added to the network without a thought on the amount of traffic that they are likely to generate, and its effect on other users. It is thus important for Network Managers to have a short-term, a medium-term and a long-term plan for the network.

            So, what are the basic elements of a network. I would say:

 

•         IP addresses/Domain names (but only if the network connects to the Internet or uses TCP/IP).

•         A network operating system (such as Microsoft Windows, Novell NetWare, UNIX and Linux). Many companies run more than one type of network operating system, which causes many problems, but has the advantage of being able to migrate from one network operating system to another. One type of network operating system can also have advantages over other types. For example, UNIX is a very robust networking operating system which has good network security and directly supports TCP/IP for all network traffic.

•         The cables (twisted-pair/fiber optic or coaxial cables). These directly affect the bit rate of the network, its reliability and the ease of upgrade of the network.

•         Network servers, client/server connections and peer-to-peer connections.

•         Bridges, routers and repeaters. These help to isolate traffic from one network segment to another. Routers and bridges are always a good long-term investment and help to isolate network traffic and can also isolate segment faults.

 

The networking topology of the future is likely to evolve around a client/server architecture. With this, server machines run special programs which wait for connections from client machines. These server programs typically respond to networked applications, such as electronic mail, WWW, file transfer, remote login, date/time servers, and so on.

            Many application programs are currently run over local area networks, but in the future many could be run over wide area networks, or even over the Internet. This means that computers would require the minimum amount of configuration and allows the standardizations of programs at a single point (this also helps with bug fixes and updates). There may also be a time when software licensing is charged by the amount of time that a user actually uses the package. This requires applications to be run from a central source (the server).

Ring-based networks have always out-performed contention-based networks (such as Ethernet), but they suffer from many problems, especially in adding and deleting stations to/from the ring, finding faults, and in start-up and shutting-down the ring. These have partly been overcome with MAUs, but the ring still needs a great deal of high-level management. Luckily, the FDDI standard has overcome the main problems.

Until recently, it seemed unlikely that Ethernet would survive as a provider of network backbones and for campus networks, and its domain would stay, in the short-term, with connections to local computers. The world seemed distended for the global domination of ATM, the true integrator of real-time and non real-time data. This was due to Ethernet’s lack of support for real-time traffic and that it does not cope well with traffic rates that approach the maximum bandwidth of a segment (as the number of collisions increases with the amount of traffic on a segment). ATM seemed to be the logical choice as it analyses the type of data being transmitted and reserves a route for the given quality of service. It looked as if ATM would migrate down from large-scale networks to the connection of computers, telephones, and all types analogue/digital communications equipment. But, remember, not always the best technological solution wins the battle for the market, a specialist is normally always trumped by a good all-rounder.

            Ethernet also does not provide for quality of service and requires other higher-level protocols, such as IEEE 802.1p. These disadvantages, though, are often outweighed by its simplicity, its upgradeability, its reliability and its compatibility. One way to overcome the contention problem is to provide a large enough bandwidth so that the network is not swamped by sources which burst data onto the network. For this, the Gigabit Ethernet standard is likely to be the best solution for most networks.

 

            Another advantage that Ethernet has over ATM is in its lack of complexity. Ethernet is simple and well supported by most software and hardware companies. ATM, on the other hand, still has many issues to be resolved, such as routing, ATM addressing and the methods of analysing and providing a given quality of service.

ISDN is a good short-term fix for transmitting digital data into an integrated digital network. Unfortunately, it creates a circuit-switched connection which means that users are charged for the amount of time that they are connected, and not the amount of data that they transfer. Is ATM the future?

Good old port 80. Without it, and TCP/IP we would not have had the WWW. The WWW has grown over the past few years. In this short time, WWW pages have evolved from pages which contained a few underlined hypertext links to complex pages which contain state-of-art graphics, animations, menus and even sound files. HTML led to JavaScript, which has led to Java, which provides a platform-independent method of running programs. So what’s next? More, and better, Java, probably.

            The tools for developing WWW pages are still not perfect but they are evolving fast. The design of many computer packages is now based on hypertext design and many use a browser-like layout (such as with navigation buttons and hypertext links). For example, documentation is typically distributed using HTML pages.

To be able to protect a network, either from misuse or from external hackers, it is important to understand the methods that users and hackers use to damage a network.

            A few words of advice for someone thinking of hacking into a network:

 

‘Don’t do it’

 

as you will probably get caught and risk being thrown off your course or sacked from your work. It happened at Intel when a part-time System Administrator ran a program which tested the security of user passwords. He was immediately sacked when it was found out that he had run the program (as he had no authorisation to run it).

            Hacking into a network can be just as damaging in term of money, as someone doing physical damage to a building or a computer system. In fact, many companies see it as a more serious crime (most buildings and computer hardware are insured and easily rebuild or repurchased, but corrupted data or misconfigured systems can take a lot of time and energy to repair)

            Another thing to remember that a hint of hacking can lead to accusations of much more serious damage. For example, a user could determine the password of another user. If at the same time there is a serious error on the other user’s data then the hacker may be accused of actually causing the damage. I have seen many occurrences of this type of situation.

            Remember. You’re a professional. Act like one.

Public-key encryption methods are the key to the security of the Internet, and any information that is transmitted over the Internet. It so easy to change for a user to change public and private keys so that the transmitted data is more secure, and then to publicise the new public-key. Many governments are obviously against this move as it allows for the total security of many types of data. In fact, several governments have banned the usage of widely available public-key source code. So who is right? I think I know.

            In the future, many forms of digital information will be encrypted. This could include telephone conversations, video conferencing, WWW accesses, remote login, file transfer, electronic commerce, running licensed applications, and so on.

Anyone who believes that a written signature is a good method of security is wrong. I have seen many occurrences of people forging another person’s signature. Also, modern electronic scanning and near-perfect quality printing allows for easy forgeries. For example, it is extremely easy to scan-in a whole document, convert it to text (using optical character recognition) and then to reprint a changed version. Thus, digital authentication is the only real way to authenticate the sender and the contents of a message. But, unfortunately, the legal system tends to take a long time to catch up with technology, but it will happen someday.

On a personal note, I have run a WWW site for many years and have seen how some users have abused their privileges, both by putting objectionable material on a WWW site and also in the material that they have accessed. One user, who held a responsible Post-Graduate position, put pages on the WWW which displayed animals and various world leaders being shot (mainly in the head). These pages had no links from the user’s home page, but could be accessed directly with the correct URL. The only way that these pages were traced was that the WWW server ran a program which logged the number of accesses of all WWW pages, each day. Day after day the accesses to the hidden index page showed that it was being accessed at least ten times more than any other page, and these were from browsers all over the world. The URL of the page had obviously been lodged with some Internet database or a weird cult group, and searches for this type of offensive page brought users to it from all corners of the planet. Immediately when it was found the user was warned and he withdrew the pages. From this, the Departmental Management Team became very wary of users adding their own personal pages, and a ruling was made that all Post-Graduate students required permission from their Research Group Leader, or Course Co-ordinator, before they could add anything to the WWW server. A classic case of one person spoiling it for the rest. Often people with the most responsible for facilities (such as Managing Directors, Head of Divisions, and so on) are not technically able to access the formal risks in any system, and will typical bar accesses, rather than risk any legal problems.

             Legal laws on Internet access will take some time to catch-up with the growth in technology, so in many cases it is the moral responsibility of site managers and division leaders to try and reduce the amount of objectionable material. If users want to download objection material or set-up their own WWW server with offensive material, they should do this at home and not at work or within an educational establishment. Often commercial organisations are strongly protected site, but in open-systems, such as Schools and Universities, it is often difficult to protect against these problems.

These days, viruses tend to be more annoying than dangerous. In fact, they tend to be more embarrassing than annoying, because a macro virus sent to a business colleague can often lead to a great deal of embarrassment (and a great deal of loss of business confidence).

            A particularity-worrying source of viruses is the Internet, either through the running of programs over Internet, the spread of worms or the spreading of macro viruses in electronic mail attachments. The best way to avoid getting these viruses is to purchase an up-to-date virus scanner which checks all Internet programs and documents when they are loaded.

            So, with viruses and worms, it is better to be safe than sorry. Let the flu be the only virus you catch, and the only worms you have to deal with are the ones in your garden.

 

© W. Buchanan, 2000