COMPUTER HARDWARE

Detailed Chapter Outline
    3.1  The Microprocessor
        3.1.1  The Central Processing Unit
        3.1.2  Microprocessor Speed
        3.1.3  Trends in Microprocessor Technology
        3.1.4  Microprocessor Architecture
         3.1.5  Microcontrollers
         3.1.6  Microperipherals

    3.2  Computer Memory
        3.2.1  Storage Hierarchy
        3.2.2  Primary Storage
        3.2.3  Secondary Storage

    3.3  Generations of Computer Hardware
        3.3.1  First Generation of Computers
        3.3.2  Second Generation of Computers
        3.3.3  Third Generation of Computers
        3.3.4  Fourth Generation of Computers
        3.3.5  Massively Parallel Computers
        3.3.6  DNA Computers

    3.4  Computer Hierarchy
        3.4.1  Supercomputers
        3.4.2  Mainframe computers
        3.4.3  Minicomputers
        3.4.4  Workstations
        3.4.5  Microcomputers
        3.4.6  Laptop or notebook computers
        3.4.7  Palmtops
        3.4.8  Personal digital assistant (PDA)
        3.4.9  Computing devices

    3.5  Input/Output Devices
        3.5.1  Human Data Entry Devices
        3.5.2  Source Data Automation Devices
        3.5.3  Output Devices
        3.5.4  Multifunction Devices

    3.6  Multimedia
 

The Problem With Hardware

Business Problem

    The Federal Aviation Administration (FAA) has a huge problem.  The agency is trying to keep 30 ancient IBM 3083 mainframe computers from suffering year 2000 failures.  According to IBM, fewer than 100 of those old mainframes are still in use.  The 3083s, manufactured in the early 1980s, have only the processing power of modern high-end desktop personal computers.
    The 3083s are used in 15 of the 20 FAA control centers.  The FAA’s five busiest centers – Fort Worth, Cleveland, Chicago, New York, and Washington – use modern IBM mainframe computers.  FAA centers keep track of all aircraft in U.S. airspace.  The 3083s take radar information and translate it into visual display data, reporting an aircraft’s location, identity, altitude, speed, and destination for air traffic controllers.
    IBM maintains that businesses are foolish to continue running applications, particularly mission-critical applications, on old mainframes.  In fact, IBM is pressuring the FAA to upgrade its mainframes to newer systems.  IBM says that it has neither the replacement parts nor people with the skills necessary to make the repairs.

Potential Business Solution

    Buying the hardware for a new system would not be very expensive.  IBM RS/6000 workstations probably have the processing power to run the application.  The larger costs would come from upgrading to a new operating system.  Analysts estimate those costs could run from $6 million to $30 million for the entire system.
 (Source:  “IBM Wants FAA to Retire 3083s.”  Computerworld, January 19, 1998)
 Question:  What should the FAA do?
 Question:  Given that the main expense is software, not hardware, why is the FAA “hanging on” to old mainframes?
 Question:  What implications would you draw for any type of computer hardware (i.e., minicomputers, workstations, personal computers, laptops, etc.)?
 

Overview
    Computers have spread throughout organizations and are used in every functional area.  Organizations and their employees use computers of every size and speed to accomplish their strategies and do their jobs more effectively and efficiently.

THE MICROPROCESSOR (CENTRAL PROCESSING UNIT OR CHIP)
 
    As we have noted in Chapter One, computer based information systems (CBIS) are composed of hardware, software, databases, telecommunications, people, and procedures.  These components are organized to input, process, and output data and information.  This chapter focuses on the hardware component of a CBIS.  Hardware refers to the actual equipment used for the input, processing, output, and storage activities of a computer system and consists of a central processing unit, input devices, output devices, primary storage, secondary storage, and communications devices (communications devices are covered in chapter 6.)
    Many companies are placing the responsibility and authority for purchasing hardware with users in the functional areas.  The management information systems group assists the users in these decisions by acting as technical advisers.  You have an excellent chance, sometime in your career, of making hardware purchasing decisions.  Therefore, a basic understanding of computer hardware is necessary and will be helpful.

The Central Processing Unit
    The central processing unit (CPU) is also referred to as the microprocessor or the chip.  Chips are composed of millions of transistors embedded on a silicon wafer.  Memory, logic, and control are contained on a single chip.
    The CPU is the center of all computer processing activities, where all processing is controlled, all data are manipulated, arithmetic computations are performed, and logical comparisons are made.  The CPU consists of the control unit, the arithmetic-logic unit (ALU), and the registers.

    Buses.  Instructions and data move between computer subsystems and the processor via communications channels called buses.  A bus is a shared data path that connects different parts of a computer system.  Bus capacities are usually given by the number of bits they carry at one time (the most common personal computer bus is 32 bits).

    Control Unit.  The control unit reads instructions and directs the other components of the computer system to perform the functions required by the program.  It interprets and carries out instructions contained in computer programs, selecting program statements from primary storage or memory, moving them to the instruction registers in the control unit, and then carrying them out.  It controls input and output devices and data-transfer processes from and to memory.  The control unit does not actually change or create data; it merely directs the data flow within the CPU.  The control unit can process millions of instructions per second, but only one instruction at a time.
    The series of operations required to process a single machine instruction is called a machine cycle.  Each machine cycle consists of the instruction cycle (steps 1 and 2 below), which sets up circuitry to perform a required operation, and the execution cycle (steps 3 and 4 below), during which the operation is actually carried out (see Figure 2).   The time to perform the instruction cycle is called the instruction time and the time to perform the execution cycle is called the execution time.
Step 1:  Fetch instruction:  The control unit gets the instruction to be executed from memory.
Step 2:  Decode instruction:  The control unit decodes the instruction so the central processor can understand what is required, moves necessary data from memory to the registers, and identifies the location of the next instruction.
Step 3:  Execute the instruction:  The arithmetic-logic unit performs what it is instructed to do, either an arithmetic computation or a logical comparison.
Step 4:  Store the results:  The results are stored in registers or memory.
    When both cycles have been completed for one instruction, they are performed for subsequent instructions.  Pipelining allows central processing units to speed up processing.  With pipelining, the CPU gets one instruction, decodes another, and executes a third simultaneously.  The Pentium processor uses two execution unit pipelines, giving the CPU the ability to execute two instructions in a single machine cycle.

Arithmetic-Logic Unit.  The arithmetic-logic unit (ALU) performs required arithmetic and comparison, or logic, operations.  The ALU adds, subtracts, multiplies, divides, compares, and determines whether a number is positive, negative, or zero.  Comparison operations determine whether a number is greater or less than another number, or if they are equal.  All computer applications are achieved through these six simple operations.  The ALU operations are performed sequentially, based on instructions from the control unit.  For these operations to be performed, the data must first be moved from primary storage to the arithmetic registers in the ALU.   Registers are specialized, high-speed memory areas for storing temporary results of ALU operations as well as for storing certain control information.

Microprocessor Speed
    The speed of a chip depends on five things:  the machine cycle time, the clock speed, the word length, the data bus width, and the design of the chip.  The machine cycle time is measured in fractions of a second.  Cycle times range from milliseconds (one-thousandth of a second) and microseconds (one-millionth of a second) for slower computers, to nanoseconds (one-billionth of a second) and picoseconds (one-trillionth of a second) for faster computers.  Machine cycle time can also be measured in terms of how many instructions are executed in one second.  This measure, called MIPS (millions of instructions per second), is used to measure speed for computers of all sizes.
    The clock, located within the control unit, is the component that provides the timing for all processor operations.  The control unit follows predetermined internal instructions, called microcode.  Microcode consists of predefined, elementary circuits and logical operations that the processor performs when it executes an instruction.  The control unit executes the microcode in accordance with the electronic cycle or pulses of the CPU clock.
    The beat frequency of the clock (measured in megahertz [MHz] or millions of cycles per second) determines how many times per second the processor performs operations.  All things being equal, a processor that uses a 400 MHz clock operates at twice the speed of one that uses a 200 MHz clock.  The number of microcode instructions needed to execute a single program instruction (e.g., performing an arithmetic operation) varies, so there is no definite, direct relationship between clock speed measured in megahertz and processing speed measures such as MIPs.
    The word length is the number of bits that can be processed at one time by a chip.  Chips are commonly labeled as 8-bit, 16-bit, and 32-bit devices.  A 32-bit chip can process 32 bits of data in a single cycle.  The larger the word length, the faster the chip speed and the greater the precision in calculations.
    The width of the buses determines how much data can be moved at one time.  The wider the data bus (e.g., 32 bits), the faster the chip.  Matching the CPU to its buses can affect performance significantly.  In some personal computers, the CPU is capable of handling 32 bits at a time, but the buses are only 16 bits wide.  In this case, the CPU must send and receive each 32-bit word in two 16-bit chunks, one at a time.  This process makes data transmission times twice as long.
If word lengths and bus widths are the same, then the larger the word length, the more powerful the computer.  Computers with larger word lengths can transfer more data in the same machine cycle.  They can also use the larger number of bits to address more memory locations.
    The design of the CPU components also helps to determine how fast the processor can operate.  The size of the integrated circuit transistors on a chip directly affects the speed at which a transistor can switch from one state to another; all things being equal, the smaller the transistor, the faster it can switch.  Also, the more densely packed the transistors, the shorter the distance electrons must travel; the shorter the distance, the faster the chip.
    All five factors (machine cycle time, clock speed, word length, data bus width, and chip design) affect the processing speed of the CPU.  Therefore, it is difficult to compare the speed of different processors.  For this reason, Intel, which controls over 75 percent of the processor market for personal computers, has created a measure of relative performance for its processors, called the iCOMP Index.  This index reflects the approximate, relative performance of Intel microprocessors on 32-bit applications.  A higher iCOMP rating means a higher relative performance of the microprocessor.  Table 1 shows the iCOMP ratings of current Intel microprocessors.

     Table 1:  iCOMP Rating of Intel Microprocessors
 
      Processor                  Clock Speed       iCOMP 2.0

        Pentium II                    400 MHz               440
        Pentium II                    350 MHz               386
        Pentium II                    333 MHz               366
        Pentium II                    300 MHz               332
        Pentium II                    266 MHz               303
        Pentium II                    233 MHz               267
        Pentium Pro (MMX)   233 MHz                203

 (Source:  www.intel.com)

Trends in Microprocessor Technology

    Table 3 shows the evolution of the microprocessor from the introduction of the 4004 in 1971 to today’s Pentium II and P6 microprocessors.  Over time, microprocessors have become dramatically faster, more complex, and more dense, with increasing numbers of transistors embedded in the silicon wafer.

                    Table 3:  Evolution of Microprocessors

Chip                 Introduction     Clock                  Bus      Number of         Addressable
                           Date              Speed                  Width     Transistors          Memory

4004                      11/71        108 KHz                 4 bits          2,300              640 bytes
8008                      4/72          108 KHz                 8 bits          3,500              16 Kbytes
8080                      4/74          2 MHz                     8 bits          6,000              64 Kbytes
8086                      6/78          5-10 MHz              16 bits         29,000              1 Mbyte
80286                    2/82          8-12 MHz              16 bits       134,000              16 Mbytes
80386 DX             10/85        16-33 MHz             32 bits       275,000               4 Gbytes
80386 SX              6/88          16-20 MHz            16 bits       275,000               4 Gbytes
80486 DX             4/89          25-50 MHz             32 bits        1.2 M                 4 Gbytes
80486 SX              4/91          16-33 MHz            32 bits        1.185 M              4 Gbytes
80586 (Pentium)    3/93          60-166 MHz           32 bits        3.1 M                  4 Gbytes
Pentium Pro           3/95          150-200 MHz         32 bits        5.5 M                  4 Gbytes
Pentium II             1996          233-300 MHz         32 bits        5.5 M                  4 Gbytes
80686                   1997         up to 400 MHz        32 bits        7.5 M                  4 Gbytes
Merced (P7)         1998         up to 500 MHz        64 bits          9 M                  16 Gbytes
80886                   2000         over 1000 MHz (estimated)

Source:  www.intel.com

    In 1970, the distance between transistors (line width) was about 12 microns (12 millionths of a meter), or about one-tenth the width of a single human hair.  By 1990, the line width had decreased to about 0.8 microns.  The latest generation of chip has 7.5 million transistors embedded on the silicon wafer, with the distances between transistors decreasing to 0.3 microns.  Recently however, scientists at the University of Texas at Austin have produced an experimental circuit with distances between circuit lines of just 0.08 microns. (see Figure 3)
    The Merced chip (or P7), being jointly developed by HP and Intel, will be in production in 1999.  The line width on this chip will be .18 microns, its clock speed is predicted to be over 500 MHz, and it is expected to contain over 50 million transistors. (see Figure 4)
    Improvements in processor manufacturing technology have led to improved processor performance.  Currently, processor performance doubles about every 18 months, a trend that was predicted by Gordon Moore, Intel Corporation cofounder.  In April 1965, Moore predicted that integrated circuit complexity would double approximately every two years.  Moore’s prediction has proven to be so accurate that it is now known as Moore’s Law.  Currently, physicists see no reason why this trend cannot continue for several more years. (see Figure 5)
    As transistors are packed closer together and the physical limits of silicon are approached, scientists are developing new technologies that increase the processing power of chips. (See Figure 5.3, Turban 1e).  Note: extend figure to include Pentium II, P6, and Merced chips with information from Table 3).  Chips are now being manufactured from  gallium arsenide (GaAs), a semiconductor material much faster than silicon because electrons can move through GaAs five times faster than they can move through silicon.  Also, GaAs chips require less power than silicon chips, but are more difficult to produce, resulting in higher prices.  GaAs chips are now being used in cellular telephones, cable TV equipment, and digital TV.
    Silicon germanium (SiGe) is a new material used in making chips that results in increased speed, reduced electronic noise, reduced expense, and reduced power requirements.  SiGe technology is promising for reducing the cost of consumer products (e.g., cellular telephones and direct broadcast satellite entertainment services), improving business applications (e.g., telephone network transmission), and helping make possible new applications (e.g., collision-avoidance automobile radar).

Microprocessor Architecture

    The arrangement of components and their interactions is called an architecture.  Computer architectures include the instruction set and the number of the processors, the structure of the internal buses, the use of caches, and the types and arrangements of input/output (I/O) device interfaces.
    Every processor comes with a unique set of operational codes or commands that represent the computer’s instruction set.  Today, two instruction set strategies, complex instruction set computing (CISC) and reduced instruction set computing (RISC), dominate the processor instruction sets of computer architectures.  These two strategies differ by the number of operations available and how and when instructions are moved into memory.
    Complex instruction set computing.  A CISC processor contains more than 200 unique coded commands, one for virtually every type of operation.  Therefore, inexpensive hardware can replace expensive software, reducing the cost of developing software.  The penalty for this ease of programming is that CISC processor-based computers have increased complexity and slower performance.  In spite of these drawbacks, most computers still use CISC processors.
    Reduced instruction set computing.  The other, more recent, approach is RISC processors, which eliminate many of the little-used codes found in the complex instruction set.  Underlying RISC design is the claim that a very small subset of instructions accounts for a very large percentage of all instructions executed.  The instruction set, therefore, should be designed around a few simple “hard-wired” instructions that can be executed very quickly.  The rest of the needed instructions can be created in software.  These instructions will be somewhat slower than built-in instructions, but they are typically used so seldom that it does not matter.  RISC processor-based computers (e.g., workstations) are significantly faster than CISC processors.
    The PowerPC chip is a RISC processor created by Motorola under agreement with IBM and Apple Computer.  Digital Equipment’s Alpha chip and Sun Microsystems’ Sparc chip are also examples of RISC processors.
    Intel has incorporated MMX (multimedia extension) technology in its Pentium microprocessors.  MMX technology improves video, audio, graphical, and image processing, encryption, and input/output processing, all of which are used in modern office software suites and advanced business media, communications, and the Internet.
    It is also possible to reduce the number of instructions in a chip by making each instruction very long.  This approach, called the very long instruction word (VLIW), is being developed by Intel and Hewlett-Packard.

Microcontrollers
    Microcontrollers are integrated circuit chips that are used in embedded applications such as automobile engine control, timing in microwave ovens, and in medical products including hearing aids, fetal monitors, and pacemakers.  Microcontrollers are also contained in computer peripherals, such as laser printers and disk drives, and are becoming increasingly important in telecommunications devices such as cellular phones.
    In fact, microcontrollers are being used throughout our automobiles.  Mechanical components such as steering columns, brakes, and throttle cables are being replaced by wires and computer chips.  The results are steer-by-wire, brake-by-wire, and throttle-by-wire systems.  In “by-wire” systems, the actions of the driver (turning the steering wheel or pressing on the brake pedal) are detected by sensors and relayed to computer chips.  These chips then signal motors to turn the wheels or activate the brakes.
    The microcontrollers can do more than just relay the driver’s intent – they can embellish that intent with intelligence.  Antilock braking systems were an early example.  Current systems are much more sophisticated.  For example, new cruise controls do not just maintain a set speed.  They use radar to gauge the speed of a slower-moving vehicle ahead and automatically slow your car to the same speed until you decide to pass.  Other sensors surround your car with an early-warning system.  Infrared detectors can spot the heat of a car in your blind spot and sound an alarm as you begin to change lanes.  New radar sensors will see past a truck ahead – by bouncing a signal off the pavement under the truck – to determine if it is safe for you to pass.

Microperipherals
    Microperipherals, such as digital signal processors (DSPs), are integrated circuit chips that perform the digital-to-analog and analog-to-digital conversions used in audio, communications, and image manipulation.  Many new applications for image manipulation are emerging, including:
    consumer photography – new “smart sensors will automatically correct exposure and allow images to be edited right on the camera;
    computers – “eyeball” digital cameras will be built into monitors for video mail or for verifying the user’s identity; keyboards with built-in fingerprint scanners will substitute for passwords;
    handheld devices – small digital cameras will become common accessories for cellular telephones and personal digital assistants;
    retailing – fingerprint and retinal scanners that can instantly verify someone’s identity will be widely used by retail stores, banks, and government agencies;
    medicine – imaging chips may help blind people see and will be commonly used in surgery;
    automobiles – digital cameras will supplement rearview mirrors, check tire pressure, and monitor occupants as part of advanced safety systems.
 
COMPUTER MEMORY
    Computers store data and information in the form of 0’s and 1’s (i.e., binary digits or bits).  For this reason, we say that computers are digital.  The storage capacity of computers is usually measured in terms of bytes (eight bits).

Storage Hierarchy
    As capacities have increased dramatically, a hierarchy of terms is used to describe computer storage.
    Kilo (abbreviated with the letter K) means one thousand, so a kilobyte (KB) is approximately one thousand bytes.  Actually, a kilobyte is 1,024 bytes (210 bytes).
    Mega (abbreviated with the letter M) means one million, so a megabyte (MB) is approximately one million bytes.  Actually, a megabyte is 1,048,576 bytes (1024 x 1024 bytes).  For example, if you own a computer with 32 megabytes of RAM, then the random access memory on your computer contains 33,554,432 bytes of storage.
    Giga (abbreviated with the letter G) means one billion, so a gigabyte (GB) is approximately one billion bytes.  Actually, a gigabyte is 1,073,741,824 bytes (1024 x 1024 x 1024 bytes).  For example, if you own a computer with 6.4 gigabytes of storage on the hard drive, then your secondary storage device contains 6,871,947,674 bytes of storage.  Suppose that you are using a word processor with 12-point font and normal spacing between lines.  With approximately 15 words per line, 22 lines per page, and 6 bytes per word, each page would contain about 2000 bytes.  Therefore, your 6.4 gigabyte hard drive would hold a little over 3.4 million pages of text.
    Tera (abbreviated with the letter T) means one trillion, so a terabyte (TB) is approximately one trillion bytes.  Actually, a terabyte is 1,078,036,791,296 bytes (1024 x 1024 x 1024 x 1024 bytes).

Primary Storage
    Primary storage, or main memory, has three main functions.  First, it stores all or part of the program being executed.  Second, it stores the operating system programs that manage the operation of the computer.  Third, it stores the data that are being used by the program.
    Primary storage in today’s microcomputers utilizes integrated circuits, which are interconnected layers of etched semiconductor materials forming electrical transistor memory units with “on-off” positions that direct the electrical current passing through them.  The “on-off” states of the transistors are used to establish a binary 1 or 0 for storing one binary digit, or bit.  A sufficient number of bits to represent specific characters – letters, numbers, and special symbols – is known as a byte, usually 8 bits.  Because a bit has only two states, 0 or 1, the 8 bits comprising a byte can represent any of 28, or 256, unique characters.  The character represented depends upon the bit combination or coding scheme used.  The two most commonly used coding schemes are ASCII (American National Standard Code for Information Interchange) and EBCDIC (Extended Binary Coded Decimal Interchange Code).
    Another coding scheme is unicode.  The Unicode Worldwide Character Standard is a character coding system that supports the interchange, processing, and display of the written texts of the world’s languages.  Unicode also supports classical and historical texts in many languages.
    There are four kinds of primary memory:  register, cache memory, random access memory (RAM), and read only memory (ROM).
    Registers.  Registers are part of the CPU where instructions and results are stored for very short periods of time.  They are the fastest, smallest, and most expensive type of memory.
    Cache Memory.  Many software programs are larger than the internal, primary storage (RAM) available to store them.  To get around this limitation, some programs are divided into smaller blocks, with each block loaded into RAM only when necessary.  However, depending on the program, continuously loading and unloading blocks can slow down performance considerably, especially because secondary storage is so much slower than RAM.  As a compromise, most processors use high-speed cache memory as temporary storage for the most often used blocks, RAM to store the next most often used blocks, and secondary storage for the least used blocks.  Cache memory operates at a much higher speed than conventional memory (i.e., RAM), because it reduces the number of times the program has to fetch instructions and data from RAM and secondary storage.
    Random Access Memory (RAM).  Primary storage that may be read or written to is known as random-access memory (RAM).  RAM is the area in which the CPU stores the instructions and data it is processing.  The larger the memory area, the larger the programs that can be stored and executed.
    With newer computer operating system software, more than one program may be operating at a time, each occupying a portion of RAM.  Most personal computers today need at least 16 to 32 megabytes of RAM to process multimedia applications, which combine sound, graphics, animation, and video.
    The advantage of RAM is that it is very fast in storing and retrieving any type of data, whether textual, graphical, sound, or animation-based.  Its disadvantages are that it is relatively expensive and that it is dynamic or volatile.  This volatility means that all data and programs stored in RAM are lost when the power is turned off.  To lessen this potential loss of data, many of the newer application programs perform periodic automatic “saves” of the data.
    Read-Only Memory (ROM).  Read-only memory (ROM) is that portion of primary storage that cannot be changed or erased.  ROM is nonvolatile; that is, the program instructions are continually retained within the ROM, whether power is supplied to the computer or not.  ROM is necessary to users who need to be able to restore a program or data after the computer has been turned off or, as a safeguard, to prevent a program or data from being changed.  For example, the instructions needed to start, or “boot,” a computer must not be lost when it is turned off.
    Programmable ROM (PROM) memory chips are used when it is necessary to make changes for a particular client or configuration.  A PROM chip can be programmed once (and only once) by a customer, thereby reducing the necessity of having to wait for the manufacturer to have a production run for a customized chip.  PROM chips are used where the CPU’s data and instructions do not change, but the application is specialized making the manufacture of a ROM chip too expensive.  For example, PROM chips are used to store the instructions for video games.
    Erasable PROM (EPROM) chips can be programmed in the field and can also be erased for reprogramming.  EPROM chips are usually used for device control, such as in robots, where the program may have to be changed on a routine basis.  An industrial robot, for example, might perform repetitive operations in a steel mill, making a particular piece of steel.  When it is necessary for the robot to work on a different piece of steel, the EPROM controlling the robot’s operation must be erased and reprogrammed.

Secondary Storage
    Secondary storage is separate from primary storage and the CPU, but directly connected to them.  Secondary storage stores data in a format that is compatible with data stored in primary storage, and provides the computer with vastly increased potential for storing and processing large quantities of software and data for long periods.  Primary storage is fixed in size, (although it can be readily increased), volatile, contained in memory chips, and very fast in storing and retrieving data.  On the other hand, secondary storage is not fixed in size and is nonvolatile.  Further, secondary storage can be on many different forms of media that are less expensive than primary storage, but are relatively slower than primary storage.  Secondary storage media include magnetic tape, magnetic disk, magnetic diskette, and optical.
    Magnetic tape.   Magnetic tape is kept on a large reel or in a small cartridge or cassette. Today, cartridges and cassettes are replacing reels because they are easier to use and access.  The principal advantages of magnetic tape are that it is very inexpensive, relatively stable, long-lasting, and can store very large volumes of data.  Magnetic tape, which is reusable, is used most frequently for backup or archival storage of data, systems backup and restoration, and off-site data storage for disaster recovery in mainframe and server systems.
    The main disadvantage of magnetic tape is its potentially slow response time, because it offers only sequential access.  That is, data are stored continuously on a tape which must be searched sequentially from the beginning to find the desired data.  Also, the magnetic tape itself is fragile and must be handled with care and the magnetic tape reels are labor intensive to mount and dismount.
    For mainframes and large servers, 0.5 inch cartridge tape drives provide large data storage capacities and high data transfer transfer rates.  Cartridges currently can store over 30 gigabytes of data and can exhibit data transfer speeds of up to 20 million bits per second.
    Storage productivity has been greatly enhanced by the introduction of automated tape retrieval systems, which use robots to retrieve and replace the tens of thousands of magnetic tape cartridges used by mainframes and large servers.  The cartridges are stored in structures called silos.  For example, StorageTek’s robotic cartridge library system holds up to 6,000 tape cartridges in a silo that is eight feet tall and twelve feet in diameter.
    Magnetic tape is also useful for personal computer and small server backup.  A single tape can typically accommodate a backup copy of most hard drives, a task that is difficult to accomplish using floppy disks.  The major tape formats used for personal computers and small servers are 8 mm tapes, 4 mm tapes (also known as digital audio tape or DAT), digital linear tape (DLT), and quarter-inch cartridges (QIC).
    Magnetic disks.   Magnetic disks (called hard disks) access data faster than magnetic tape by assigning specific address locations for data, so that users can go directly to the address without having to go through intervening locations looking for the right data to retrieve.  Therefore, we say that magnetic disks offer direct access to data.  Most computers today rely on hard disks for retrieving and storing large amounts of instructions and data in a nonvolatile and rapid manner.  The hard drives of modern microcomputers can provide up to 20 gigabytes of data storage.  (See Figure 6)
    Hard drives keep getting smaller and providing more storage.  IBM recently introduced a hard drive weighing less than an AA battery that can hold 340 megabytes of data.  This hard drive is used in car navigation systems to store maps and addresses, in handheld computers to store schedules and telephone numbers, and in digital cameras to provide storage for digitized photographs.
    A hard disk drive is a stack of metal-coated platters (usually permanently mounted) that rotate rapidly.  Magnetic read-write “heads”, attached to arms, hover over the platters.  To locate an address for storing or retrieving data, the head moves inward or outward to the correct position, then waits for the correct location to spin by underneath.  Most hard-disk drives are housed in hermetically sealed containers to avoid contamination.
    Over time, magnetic disks and disk drives have gotten smaller and simultaneously faster, denser, and less expensive.  According to Dataquest, the cost per megabyte of hard disk storage fell from $75 in 1982 to $0.45 today.  As the cost per megabyte has fallen, the number of megabytes per drive has increased.  Therefore, the price for a disk drive has remained relatively constant.
    The speed of access to data on hard disk drives is a function of the rotational speed of the disk and the speed of the read/write heads.  The read/write heads must position themselves, and the disk pack must rotate until the proper information is located.  Advanced disk drives rotate at 7,200 revolutions per minute and have access speeds of 1.5 to 10 milliseconds.
    Magnetic disks provide storage for large amounts of data and instructions that can be rapidly accessed.  Their disadvantage is that they are more expensive than magnetic tape and they are susceptible to “disk crashes.”  The read-write heads float just above the surface of the platters, at a distance of only 0.25 micrometers (25 millionths of a meter).  For comparison, consider that the read-write heads float above the platters at a distance smaller than a particle of smoke.  A disk crash occurs when the read-write heads come into contact with the surface of the platters, thereby damaging both the heads and the disk surfaces.  Though rare, disk crashes usually result in catastrophic loss of data.
    Another disadvantage of magnetic disks is that updating information stored on a disk destroys the old information because the old data on the disk is written over when changes are made.  In contrast, changes to data made on magnetic tape are made on a different reel of tape so that the old version of the tape can be retained and recovered.
    Removable disk drives were developed to accommodate larger storage requirements generated by the increased multimedia content of business and consumer software.  Increased business travel also created a need for reliable, portable data transfer media that could hold larger amounts of data.
    Removable disk drives are faster than removable floppy disk drives because their disks are composed of compact, rigid platters that can spin at a faster rate.  The one gigabyte Iomega Jaz cartridge, for example, spins at a rate fast enough for multimedia playback as well as storage.
    Hard drives in all computer systems are susceptible to failures caused by temperature variance, head crashes, motor failure, controller failure, and changing voltage conditions.  In addition, operating system malfunctions, viruses, and heavy input and output traffic impact disk reliability.  To improve reliability and protect the data in their mass storage systems, companies are combining a large number of small disk drives developed originally for microcomputers.  These devices are called Random Arrays of Inexpensive Disks (RAID).  Because data are stored across many drives, the overall impact on system performance is lessened when one drive malfunctions.  Also, multiple drives provide multiple data paths, improving performance.  Finally, because of manufacturing efficiencies of small drives, the cost of RAID devices is significantly lower than the cost of large disk drives of the same capacity.
    Disk mirroring (keeping an exact copy of one disk on another) is the simplest way of protecting data, but it requires twice the disk capacity and associated cost.  Mirroring, known as RAID 1, is used by many information-intensive industries such as banks and insurance companies.
    A disk controller controls the transfer of information between the disk drives and the rest of the computer system.  Most personal computers today use one of two high-performance interface standards:  the Integrated Drive Electronics (IDE) interface or the Small Computer Systems Interface (SCSI).
    Hard disk drives featuring IDE interfaces are standard in today’s personal computers because they offer good performance and are inexpensive.  A newer version of the IDE specification known as Enhanced IDE (EIDE) supports more devices, such as CD-ROMs, tape drives, and optical discs, and offers increased transfer rates.
    Hard disk drives featuring small computer system interfaces (SCSI) are generally found in servers and high-end workstations.  These interfaces are rapidly gaining acceptance in desktop computers.  SCSI drives are more expensive than IDE drives, but they can transfer data at faster rates.
    Magnetic Diskettes.  To transport data or instructions from one personal computer to another,  developers created the magnetic diskette, also called the “floppy disk.”  The floppy disk is a small, removable, flexible magnetic platter encased in a plastic housing.  Unlike the hard disk drive, the read-write head of the floppy disk drive actually touches the surface of the disk.  As a result, the speed of the floppy drive is much slower, with an accompanying reduction in data transfer rate.  However, the diskettes themselves are very inexpensive, thin enough to be shipped into books or mailed, and able to store relatively large amounts of data.
    The most common floppy disks today are the 3.5 inch rigid case diskettes with a capacity of 1.44 MB.  High-density, removable floppy disk drives are faster than standard 1.44 MB floppy drives and use disks with much higher capacities, ranging from 25 MB to more than 200 MB.  For  example, Iomega Corporation produces the 100 megabyte Zip drive, and Syquest produces the 230 megabyte EZFlyer and 4.7 gigabyte Quest models.
    Optical storage devices.  Optical storage devices use lasers that write to and read from disks.  These devices have extremely high storage density.  Typically, much more information can be stored on an optical disk than on a comparably sized magnetic disk.  Because a highly focused laser beam is used to read or write information encoded on an optical disk, the information can be highly condensed.  In addition, the amount of physical disk space needed to record an optical bit is much smaller than that usually required by magnetic media.
    Another advantage of optical storage is that the medium itself is less susceptible to contamination or deterioration.  First, the recording surfaces (on both sides of the disk) are protected by two plastic plates, which keep dust and dirt from contaminating the surface.  Second, only a laser beam of light comes in contact with the recording surface, not a flying head; the head of an optical disk drive comes no closer than 1 mm from the disk surface. Optical drives are also less fragile, and the disks themselves may easily be loaded and removed.
    Optical storage media do have disadvantages.  Optical disks are slower than magnetic disks, and optical disk drive mechanisms are more expensive than magnetic drives.
    Automated optical disc changers, known as optical jukeboxes, typically hold several discs and contain one or more drives to provide access to multi-volume libraries of data.  Optical storage is a reliable medium that falls between the less expensive, but slower tape drive options and the more expensive, but faster magnetic disk drive options.
    Three common types of optical drive technologies are CD-ROM, WORM, and rewritable optical.  Compact disk read-only memory (CD-ROM) disks have high capacity and low cost.  CD-ROM technology is very effective and efficient for mass producing many copies of large amounts of information that does not need to be changed.  For example, encyclopedias, directories, and on-line databases use CD-ROM technology.
    After a master disc has been produced, copies can be manufactured very inexpensively.  The unit cost is somewhat less than $1.00 for quantities of 2,000 discs.  Because of this inexpensive duplication, CD-ROMs having capacities of 650 megabytes are ideal for publishing large amounts of text, data, computer programs, or images in an electronically readable form.
    CD-ROM discs have almost completely replaced magnetic tape and magnetic disks for the distribution of software and documentation for personal computers and workstations.  In addition, CD-ROM drives are now becoming standard on new personal computer systems.  Most CD-ROM disc drives play audio and data discs interchangeably.
    CD-ROMs are generally too expensive for unique, one-of-a-kind applications.  For these situations, “write once, read many” (WORM) technology is more practical.  WORM drives are useful for archiving digitized documents.  The storage capacity and access time for WORMs are virtually the same as those with CD-ROMs.
    Advances in CD-ROM technology include CD-Rewritable and digital video discs (DVD).  When information needs to be changed or updated, and companies do not want to incur the expense of mastering and producing a new CD-ROM, rewritable optical disks are needed.
    Digital video disks offer the storage equivalent of up to about eight CD-ROM discs.  With advanced compression technologies, one digital video disk will hold a 133-minute movie accompanied by a surround-sound audio track.
    The DVD standard specifies that players will be able to read current CDs and CD-ROMs, as well as DVDs.  Two sided DVDs will be able to hold up to 17 gigabytes.  Like CD-ROMs, DVDs will be used to store music, movies, and multimedia packages.  However, DVD packages will be enhanced with features, that, for example, might offer viewers a choice of camera angles or several different sound track languages.  Experts predict that in the near future, DVD technology will produce a product that could hold more than 50 GB on a 1.2 millimeter thick platter, the equivalent of a small library on a single disc.
    Flash memory.   Flash memory, a type of nonvolatile memory, can be reprogrammed. It is either built into a system or installed on a personal computer card (known as a flash card, which has 40 or 80 megabytes).  These cards consume less space and less power than a disk drive, and are very important in small, hand-held battery-powered devices such as portable telephones and digital cameras.  Figure 7 shows the relationships among primary and secondary storage devices along the dimensions of cost, size, and speed.

GENERATIONS OF COMPUTER HARDWARE
    Computer hardware has evolved through four stages, or generations, of technology.  Each generation has provided increased processing power and storage capacity, while simultaneously exhibiting decreases in costs.  For example, the cost of performing 100,000 calculations decreased from several dollars in the 1950s to less than $0.025 in the 1980s to about $.00004 in 1995.  The advent of new technologies that perform the processing functions has dictated the beginning of each successive generation of computers.

First Generation of Computers
    The first generation of computers (circa 1946-1956), used vacuum tubes to store and process information.  Vacuum tubes consumed large amounts of power, generated much heat, and were short-lived.  First generation computers had limited memory and processing capability.  The size of main memory was about two kilobytes and the computer speed was approximately ten thousand instructions per second.  Rotating magnetic drums were used for internal storage and punched cards for external storage and data and program input.  Running programs and printing output were coordinated manually.

Second Generation of Computers
    The second generation of computers (circa 1957-1963), used transistors for storing and processing information.  Transistors consumed less power than vacuum tubes, produced less heat, and were cheaper, more stable, and more reliable.  For memory, these computers used magnetic cores, composed of small, magnetic doughnuts that could be polarized in one of two directions to represent one bit of data.  These cores were strung with wire to read and write data.  This system had to be assembled by hand, and was very expensive.  These computers had up to 32 kilobytes of RAM and speeds of up to 300 thousand instructions per second.  Second-generation computers, with increased processing and storage capabilities, began to be more widely used for scientific and business purposes.

Third Generation of Computers
    Third-generation computers (circa 1964-1979), used integrated circuits for storing and processing information.  Integrated circuits are made by printing numerous, small transistors on silicon chips.  These devices were called semiconductors.  Computer memories expanded to two megabytes of RAM and speeds increased to five MIPs.  Third-generation computers employed software that could be used by non-technical people, thus enlarging the computer’s role in business.

Fourth Generation of Computers
    Fourth-generation computers (circa 1980-present), use very large scale integrated circuits to store and process information.  Computer memory has increased to over two gigabytes in large machines and processing speeds can exceed 1000 MIPs.  These computers are inexpensive and are widely used in business and everyday life.
    The first four generations of computer hardware are based on the Von Neumann architecture, which processes information sequentially, one instruction at a time.  The fifth generation of computers uses massively parallel processing to process multiple instructions simultaneously.

Fifth Generation:  Massively Parallel Computers
    Normal desktop computers have a single processor.  However, some computers have more than one processor and can perform multiprocessing.
    Coprocessing.  A simple form of multiprocessing is coprocessing.  A coprocessor is optimized for specific tasks, such as computations or graphics.  The coprocessor speeds processing by executing specific types of instructions while the processor works on other instructions or activities.
    Parallel Processing.  Another form of multiprocessing is parallel processing, in which computers have multiple processors.  Massively parallel computers use flexibly connected networks of microprocessors.  As opposed to parallel processing, where small numbers of powerful but expensive microprocessors are linked together, massively parallel machines link thousands of inexpensive, commonly used chips to address large computing problems, attaining supercomputer speeds.  With this type of machine, business problems are broken into many parts.  Each part is assigned to a processor and the results are reassembled to obtain a final solution.
    With enough chips networked together, massively parallel machines can perform more than a trillion floating point operations per second (teraflops).  A floating point operation is a basic computer arithmetic operation, such as addition or subtraction, on numbers that include a decimal point.
    Parallel processing architectures can be divided into two broad areas:  single instruction/multiple data (SIMD) and multiple instruction/multiple data (MIMD).  SIMD computers execute the same instruction on many data values simultaneously.  MIMD connects a number of processors that run different programs or parts of programs on different data sets.  MIMD computers can further be subdivided into shared-memory and distributed-memory machines.
    In shared-memory MIMD computers, all processors access a common memory via a shared high-speed bus.  Communications among processors is easy because each processor leaves its answer in memory and tells the other processors the address in memory at which to find the answer.  The disadvantage is that many processors must share the same bus, making it possible that processors must wait to use the bus.
    In distributed-memory MIMD systems, each processor has its own memory store and communicates via high-speed buses.  That is, every computing node is a complete computer, each with its own local memory.  Because nodes cannot access other nodes’ private memory, results must be passed among nodes over a communication network.  With distributed-memory MIMD computers, the speed of the network is the limiting factor.
    Scalability is the biggest advantage of distributed-memory systems over shared-memory systems. As shared-memory machines add processors, the overall system performance may degrade due to bus contention.  As processors are added in distributed-memory computers, the processing power increases linearly.  The biggest disadvantage of distributed-memory machines is that they are difficult to program.

DNA Computers
    A recent advance in computing is called DNA computing, which takes advantage of the fact that information can be written onto individual DNA molecules.  This information uses the alphabet of four bases that all living organisms use to record genetic information.  DNA computations code a problem into this alphabet and then create conditions under which DNA molecules are formed that encode all possible solutions of a problem.  The process produces billions of molecules encoding wrong answers, and a few encoding the right one.  Modern molecular genetics has chemical procedures that can reliably isolate the few DNA molecules encoding the correct answer from all the others.
    Although DNA computers are useful for only a small number of problem types, they do process in parallel and are potentially twice as fast as today’s fastest supercomputers.  They also have the potential for storage capacities one trillion times greater than current storage media.

COMPUTER HIERARCHY
    Computers may also be distinguished on the basis of their processing capabilities.  Computers with the most processing power are also the largest and most expensive.

Supercomputers
    Supercomputers are the computers with the most processing power.  The primary application of supercomputers has been in scientific and military work, but their use is growing rapidly in business as their prices decrease.  Supercomputers are especially valuable for large simulation models of real-world phenomena, where complex mathematical representations and calculations are required, or for image creation and processing.  Supercomputers are used to model the weather for better weather prediction, to test weapons nondestructively, to design aircraft (e.g., the Boeing 777) for more efficient and less costly production, and to make many sequences in motion pictures (e.g., “Star Wars” and “Jurassic Park”).  Supercomputers generally operate at 4 to 10 times faster than the next most powerful computer class, the mainframe.
    Cray is a leading supercomputer manufacturer.  Many firms use Cray supercomputers.  For example, Nissan became the first Cray user in the Japanese automobile industry in 1986.  Nissan reports that Cray systems have played a significant role in improving quality, reducing costs, and shortening development time in their automotive development processes.

                         Supercomputers and Weather Forecasting
 
    In April 1997, the National Oceanic and Atmospheric Administration (NOAA) moved to enhance the United States’ weather forecasting capability.  NOAA announced that it would provide $18 million over three years to establish the International Research Institute (IRI), which is dedicated to providing early warnings of climate variability that influences drought, floods, and other destructive weather patterns around the world.  IRI scientists rely heavily on Cray supercomputers and Silicon Graphics workstations to produce, manipulate, and analyze climate models.
    To determine how the weather will likely affect various global regions, the scientists use output from three coupled ocean-atmosphere computer models to forecast sea-surface temperatures.  Each of these models takes into account the sea-surface temperature variations over history and each processes the information using slightly different calculations for various weather factors.  The scientists statistically “blend” the models to get a consensus forecast.  The scientists then use a Cray supercomputer to run several general atmospheric circulation models, which rely in part on information contained in the sea-surface temperature models to produce complete forecasts for different regions of the world.
(Source:  www.cray.com/products)
    Question:  Besides sea-surface temperature variations, what other variables would you include in a model to predict the weather?
    Question:  It would be cheaper to use a mainframe for weather prediction models.  Why do the IRI scientists need a supercomputer?

Mainframe Computers
    Mainframes are not as powerful and generally not as expensive as supercomputers.  Large corporations use mainframe computers for centralized data processing and maintaining large databases.  Applications that run on a mainframe can be large and complex, allowing for data and information to be shared throughout the organization.  Examples of mainframe applications include airline reservation systems, corporate payroll, and student grade calculation and reporting.
    A mainframe system may have anywhere from 50 megabytes to several gigabytes of primary storage.  On-line secondary storage may use high-capacity magnetic and optical storage media with capacities in the gigabyte to terabyte range.  Additionally, off-line storage often uses high-capacity magnetic tape systems.  Typically, several hundreds or thousands of on-line computers can be linked to a mainframe.
    The size, power consumption, and cost of mainframe computing have dramatically decreased.  Today’s most advanced mainframes perform at more than 1000 MIPS and can handle up to one billion transactions per day.

                     Recentralization of the Computing Resource

    Recentralization, which means moving computer-based applications back to large
servers (i.e., mainframes) from desktop machines and smaller, midrange servers, is gaining momentum at medium and large-sized firms.  The reasons for the shift include supporting the high transaction levels associated with electronic commerce, reducing the total cost of ownership of distributed systems, simplifying administration, reducing support personnel requirements, and improving system performance.  In addition, host computing is less expensive than distributed computing and provide a secure, robust computing environment in which to run strategic, mission-critical applications.
    Chrysler began consolidating manufacturing, finance, decision-support, and other applications onto Sun and IBM hosts in 1996, and has realized three benefits.  First, Chrysler has noted a decrease in total cost of ownership, resulting from lower hardware-acquisition costs, reduced maintenance, and higher availability.  Second, Chrysler has experienced improved capacity and resource utilization, allowing, for example, one server to support data inquiries during the day and batch transactions at night.  Also, several servers can share a single host storage device.  Third, the standardized environment that resulted from recentralization has reduced application development time.
    (Source:  Garvey, M. J.  “Mainframes Bounce Back.”  InformationWeek, July 7, 1997;14-15.)
    Question:  What are the disadvantages associated with recentralizing the computing resource?

Minicomputers
    Minicomputers, also called midrange computers, are relatively small, inexpensive, and compact computers that perform the same functions as mainframe computers, but to a limited extent.  Minicomputers are usually designed to accomplish specific tasks such as process control, scientific research, and engineering applications.  IBM is the market leader in minicomputers with its AS/400 series of computers.
    Larger companies gain greater corporate flexibility by distributing data processing with minicomputers in organizational units instead of centralizing computing at one location.  These minicomputers are connected to each other and often to a mainframe through telecommunication links.  The minicomputer is also able to meet the needs of smaller organizations that would rather not utilize scarce corporate resources by purchasing larger computer systems.
    For example, the city council of Parramatta, a city close to Sydney, Australia, handles hundreds of inquiries on a daily basis, from dog licenses to development applications, and parking fines to property ownership.  Recently, the council purchased an AS/400 with 64-bit RISC processing.  The AS/400 system resulted in extraordinary performance gains over its old methods of manually handling the inquiries.  Response times to customer inquiries were reduced by more than 80 percent in some instances, and batch reporting that had taken several hours with the city’s old mainframe computer was cut in half.

Workstations
    Vendors originally developed desktop engineering workstations, or workstations for short, to provide the high levels of performance demanded by these users.  Workstations are typically based on RISC (reduced instruction set computing) architecture and provide both very high-speed calculations and high-resolution graphic displays.  These computers have found widespread acceptance within the scientific community and, more recently, within the business community.
    The distinction between workstations and personal computers is rapidly blurring.  The latest personal computers have the computing power of recent workstations.  Low-end workstations are now indistinguishable from high-end personal computers.
    Physician Professional Services (PPS) is a full-service billing and information management organization.  Created and managed by physicians for physicians, the firm monitors the latest trends in reimbursement, payment methodologies, government and insurance carrier guidelines, laws, and regulations.  Its mission is to help physicians meet financial goals and effectively manage patient data so they may deliver quality patient care.  To meet these objectives, PPS uses IBM’s RISC System/6000 Workstation with data on 400,000 patient visits and about 150,000 active patient registrations.  The workstation manages 3,000 transactions, and 1000 documents, per day – charges, payments, and daily financial activity.  It has reduced backup time from 5.5 hours to two, from 23 reels of tape to one 8 mm cartridge.  Month-end closes went from three hours to 14 minutes.

Microcomputers
    Microcomputers (also called micros or personal computers) are the smallest and least expensive category of general-purpose computers.  They may be subdivided into four classifications based on their size:  desktops, laptops, notebooks, and palmtops.
    The desktop personal computer is the typical, familiar microcomputer system which has become a standard tool for business and, increasingly, the home.  It is usually modular in design, with separate but connected monitor, keyboard, and CPU.  In general, modern microcomputers have between 32 and 256 megabytes of primary storage, one 3.5 inch floppy drive, a CD-ROM drive, and several gigabytes or more of secondary storage.
    Network computers, also called thin clients, are computers that do not have the full functionality of typical desktop machines, and allow users to access a network.  It is unlikely that network computers will cause today’s personal computers to become obsolete.  Instead, NCs will become another option for the company.  NCs work best in the following situations:
            Users who work with a limited set of programs.  These users might range from secretaries, who spend the majority of    their time on word processing, to top-level executives, who primarily interact with electronic mail.
            Shared desktops.  The corporate world is moving away from the model of one person for one office.  Virtual workgroups, which may be comprised of consultants, contractors, part-time employees, and full-time employees, form regularly.  Purchasing a personal computer for every individual who might need one is not cost-effective.  Because network computers store user-specific configuration data on the server (and not locally), they can be easily shared, while still allowing each employee to enjoy his or her personalized working environment.
            Remote users who are difficult to support.  As telecommuting becomes more prevalent, the cost and difficulty of supporting remote users has grown exponentially.  As there are few components in a network computer, there is less that can go wrong and therefore less to fix.
            Whenever security is critical.  Conventional desktop and laptop personal computers store information on local hard drives and give users virtually unlimited access in local storage.  NCs store everything on the server and are generally safer because server rooms are more physically secure than the typical machine sitting on the desktop.  In addition, network managers regularly back up their servers.
    Network computers are not only less expensive than standard personal computers, but accrue additional cost-benefit over the life of the computer.  Savings can be achieved with NCs through less technical support, less training for users, and less frequent replacement because NCs do not become obsolete as quickly as personal computers.

                    Network Computers at Fred Meyer, Inc.
 
    Fred Meyer, Inc., with $15 billion in sales, has become one of the nation’s five largest food and drug retailers through strategic acquisitions of several major regional grocery retailers.  Part of the reason for Fred Meyer Inc.’s success has been its strategic use of technology in all its operations – integrated merchandising logistics and inventory management systems.  With one of the recent acquisitions, Smith’s Food and Drug Centers, a 156 store Western supermarket chain, Fred Meyer Inc. faced a difficult problem.  To deploy Fred Meyer Inc.’s pharmacy system at Smith’s, the company would have to place both terminals and personal computers in the pharmacy service counters.  This process would result in both an administrative headache for IT, and a problem for employees.
    However, Fred Meyer Inc.’s CIO had a solution – network computers.  She had compelling reasons for using NCs.  They connect to regular computer monitors and display the same personalized programs as on a personal computer, such as the Internet or Windows and mainframe applications.  But, instead of depending on built-in memory and local disk drives, which can be an entry point for viruses and incompatible software, NCs get information, such as corporate and personal files, from more reliable servers.  Rather than replacing multiple, obsolete personal computers every two years, and updating the software on everyone’s desktop every time there was an upgrade, changes need only be made to the few servers to which the NCs are connected.  This solution was much cheaper and more efficient that keeps performance, hardware, and software up to date.
    After testing the NCs in the Fred Meyer IT lab and deploying them in a pilot at a Fred Meyer store, the CIO was convinced that the NCs could provide access to the applications that Smith’s pharmacy personnel needed, namely Unix applications residing on local workstations, and mainframe applications located on a remote mainframe at headquarters.  She also felt that there would be a lower total cost of ownership than with personal computers.
    Fred Meyer Inc. is also planning to use NCs in other ways.  The company is planning to provide NCs with Lotus’ suite of productivity applications to employees who do not require personal computers.  With NCs, employees will be able to access company e-mail and the company intranet.  Instead of distributing training CD-ROMs to each store, employees will use NCs to access the latest training programs on a server at headquarters.
    (Source:  www.pc.ibm.com/networkstation).
    Question:  As a CIO, would you feel safe using only network computers?  Discuss why or why not.
    Question:  Relate the Fred Meyer case to the statement:  “The network is the computer.”

Laptop Computers and Notebook Computers
    As computers become much smaller and vastly more powerful, they become portable and new ways of using them open for users.  Laptop or notebook computers are small, easily transportable, lightweight microcomputers that fit easily into a briefcase.  Laptops and notebooks are designed for maximum convenience and transportability, allowing users to have access to processing power and data without being bound to an office environment

Palmtop Computers
    Palmtop computers are hand-held microcomputers, small enough to carry in one hand.  Although still capable of general-purpose computing, palmtops are usually configured for specific applications and limited in the number of ways they can accept user input and provide output.

Personal Digital Assistants
    A personal digital assistant (PDA) is a handheld, palmtop computer that uses a pen rather than keyboard input.  PDAs provide electronic notepad, calendar, and wireless communication capabilities.  PDAs differ from other personal computers in that they are usually specialized for individual users.  Users must train their PDAs to recognize their handwriting by writing each letter and digit several times.  The PDA may be thought of as a computing appliance, rather than a general-purpose computing device.
    To facilitate input, PDAs are designed to work with handwriting recognition as well as software that may anticipate a user’s intentions.  For example, a user with a PDA simply has to jot down “Call John” for the PDA to bring up an address book with phone numbers for all entries with the name John.  By selecting the appropriate entry, a wireless link can be used to make the connection.  Wireless connections are also used to link into a wide array of services such as electronic mail (E-mail), electronic information sources, and electronic paging.  Infrared connection to local area networks may also be accomplished with PDAs.
Pen-based input appeals to people who do not like to use keyboards (e.g., executives) or who cannot conveniently use a keyboard in their work.  For example, restaurant waiters can transmit orders directly from your table to the kitchen and then use the PDA to get your signature for a credit card charge.  Salespeople can use PDAs to provide wireless connections to their headquarters, home, or the Internet.
    At present, PDAs recognize text written in unconnected block printing, with each character printed separately in uppercase with the electronic stylus on the screen.  This recognition is only 90-95 percent accurate.  The problems of entering text with a stylus mandate alternate modes of entry.  For example, you can delete a word by crossing it out on the screen.  Tapping the stylus on the name of a stored document will bring it onto the screen.  Text can also be entered by tapping with the stylus on the appropriate letters from an on-screen virtual keyboard.
    PDAs are used on the Chicago futures exchange, where commodities are traded.  PDAs replace the paper cards that the traders used that encouraged fraud.  PDAs provide instant matching of seller to buyer as well as a crucial audit trail to make sure that the customers’ trades are properly executed.  The PDAs include a radio transmitter that instantly broadcasts the data on each trade to the exchange’s computers.
    An example of a PDA is the PalmPilot III, made by U.S. Robotics, which offers seven applications:  a date book, an address book, a to-do list, a notepad, an expense report, e-mail, and a calculator.  To exchange data between PalmPilots, users can beam information from one PalmPilot III to another using an infrared port.  The PalmPilot III requires users to learn its Graffiti pen-input alphabet.  Users can place the PalmPilot III into a docking cradle and synchronize their PalmPilots with their personal computers.

Computing Devices
    Wearable computers free their users’ movements.  An electric utility has given their technicians wearable computers to free them from having to carry cumbersome paper manuals.  The device is strapped to their belts and the display unit is positioned close to the eye.  A microphone can receive spoken commands directed at the unit or at a remote server and transmit them by wireless telecommunications.
    Wearable computers developed at Carnegie Mellon University include a position-sensing facility that continually determines the user’s location within a geographic area.  Knowing this, the computer can provide the user with information relating to his or her current position.  The military is currently performing research for combat troops.
    Media Lab at MIT has a smart-clothes project.  Microprocessors, cameras, microphones, and sensors that are built into your clothing enable the system to “know” your environment and help you as an intelligent assistant.  The goal of the project includes supplying the names of the people you meet or giving you directions to your next appointment.  If a business associate mentions an impending merger and acquisition to you, the system will “hear” and project the details of the project onto the display in your glasses.
    Embedded computers are placed inside other products to add features and capabilities.  For example, the average mid-sized automobile has over 3,000 embedded computers that monitor every function from braking to engine performance to seat controls with memory.
    Active badges have been introduced by Xerox to be worn as ID cards by employees who wish to stay in touch at all times while moving around the corporate premises.  The clip-on badge contains a microprocessor that transmits its (and its wearer’s) location to the building’s sensors, which send it to a computer.  When someone wants to contact the badge wearer, the phone closest to the person is identified automatically.  When badge wearers enter their offices, their badge identifies them and logs them on to their personal computers.
    Memory buttons are nickel-sized devices that store a small database relating to whatever it is attached to.  These devices are analogous to a bar code, but with far greater informational content and a content that is subject to change.  The U.S. Postal Service is placing memory buttons in mailboxes to improve collection schedules.
An even smaller form of computer is the smart card.  Similar in size and thickness to ordinary plastic credit cards, smart cards contain a small processor and memory that allow these “computers” to be used in everyday activities such as personal identification and banking.
    Uses for smart cards are appearing rapidly.  People are using them as checkbooks; a bank ATM (automatic teller machine) can “deposit money” into the card’s memory for “withdrawal” at retail stores.  Many states and private health maintenance organizations are issuing smart health cards that contain the owner’s complete health history, emergency data, and health insurance policy data.  Smart cards are being used to transport data between computers, replacing floppy disks.  Adding a small transmitter to a smart card can allow businesses to locate any employee and automatically route phone calls to the nearest telephone.
    Smart cards have caught on in much of the world, but not in the United States.  The reason is that consumers and other users are not sold on smart cards’ benefits.  For example, Oklahoma State University has offered smart cards for use at its sports facilities since January 1997, but sales have been slow.  Also, the public’s lack of “comfort” with smart cards could be an obstacle to widespread acceptance.  Another reason is America’s superior telecommunications infrastructure.  This infrastructure permits reliable and affordable electronic transactions, so that credit cards, with their magnetic stripes, can be easily and quickly verified with a connection to the card-issuing company.  But, in other countries, telecommunications networks are not nearly as reliable, cheap, or even available as they are in the U.S.  In those countries, smart cards work well, because they are self-contained devices and do not require connections to a central location.
    For example, Germany has distributed smart cards containing health-care information – not personal medical histories, but payment-processing information – to all its citizens.  The program has reduced claims-processing costs by 35%, so the system paid for itself in 18 months.
    Many proponents of smart cards say that what is missing in the U.S. is a business justification – the “killer application.”  Visa and Siemens Corporation may have an answer.  They are launching a test that will give Siemens’ employees a smart card for travel.  The cards will ensure that when the employees are on the road, they will pay only the rates Siemens has negotiated with selected hotels, airlines, and car-rental companies.  If the smart cards work as planned, they could help Siemens recover the $1 million a year the company says it loses because traveling employees are not always charged Siemens’ negotiated rates.
Another form of smart card has become common with laptop and palmtop computers – the Personal Computer Memory Card International Association (PCMCIA) card.  With computers that have corresponding PCMCIA connectors, these cards allow users to add capabilities such as faxing and connections to networks, to add more primary memory, or to run specialized application software.

INPUT/OUTPUT DEVICES
    The user interface is the most important component of a computer system to the user.
Input and output devices are integral parts of this human-computer interface, which includes other hardware devices and software that allow humans to interact with computer systems.  The user interface is the most important component of a computer system to the user.
    The input/output (I/O) devices of a computer are not part of the CPU, but are channels for communicating between the external environment and the CPU.  I/O devices are controlled directly by the CPU or indirectly through special processors dedicated to input and output processing.  Data and instructions are entered into the computer through input devices and processing results are provided through output devices.  There are numerous input and output devices.
    It is critically important for organizations to have fast, accurate data input into their computer systems.  If the data are not accurate, the acronym, GIGO (garbage-in-garbage-out) comes into play.  Getting data into the computer usually requires transferring human-readable data into the computer and translating these data into machine-readable data.  Human-readable data can be directly read and understood by humans, where machine-readable data can be understood and read by computer devices (e.g., the universal bar code read by scanners at grocery checkouts).  Data may be human-readable and machine-readable (e.g., magnetic ink on bank checks).
    Humans may enter data into a computer, mainly through a keyboard, mouse, or touchscreen.  This type of data entry is prone to error.  An alternative form of data entry is source data automation, which captures data in computer-readable form at the moment the data are created.  Point-of-sale systems, optical bar-code scanners, other optical character recognition devices, handwriting recognizers, voice recognizers, digitizers, and cameras are examples of source data automation.  Source data automation eliminates errors arising from human data entry and allows for data to be captured directly and immediately, with built-in error correction.

Human Data Entry Devices
    Users can command the computer and communicate with it by using one or more input devices to trigger the action language.  The action language is how the user tells the computer what to do.  Users want communication with computers to be simple, fast, and error free.  Therefore, a variety of input devices fit the needs of different individuals and applications.  In this section, we will deal with the following devices:  keyboard, mouse, touchscreen, light pen, and joy sticks.
    Keyboards.  The most common input device is the keyboard.  The keyboard is designed like a typewriter but with many additional special keys.  Most computer users utilize keyboards regularly.  However, a number of computer users have developed repetitive stress injury which they allege comes from excessive use of poorly designed keyboards.  As a result, new keyboards have been developed that are better designed ergonomically.  For example, some keyboards are now split in half with large wrist rests, more closely approximating the natural angle of the arms and wrists.
    DataHand Systems of Phoenix, Arizona has developed the DataHand keyboard, which consists of two unattached pads.  Rather than a conventional array of keys, the device has touch-sensitive receptacles (or finger wells) for the fingers and thumbs.  Each finger well allows five different commands, which are actuated by touching one of the sides or the bottom of the finger wells.  Complex commands can be programmed so that a single flick of the finger can be used to enter frequently used sequences of commands or chunks of data.
    Mouse.  The mouse is a hand-held device used to point a cursor at a desired place on the screen, such as an icon, a cell in a table, an item in a menu, or any other object.  Once the arrow is placed on an object, the user clicks a button on the mouse, instructing the computer to take some action.  The use of the mouse reduces the need to type in information or use the slower arrow keys.  Special types of mouse are rollerballs and trackballs, used in many portable computers.  A new technology called glide-and-tap allows fingertip cursor control.
    Touch Screen.  An alternative to the mouse or other screen-related devices is a touch screen.  The user activates an object on the screen by touching it with his or her finger.
    Light Pen.  A light pen is a special device with a light-sensing mechanism which is used to touch the screen.  Pointing with a light pen is more accurate than touch screens because you can point at very small objects.
    Joy Sticks.  Joy sticks are used primarily at workstations that can display dynamic graphics.  They are also used to play video games.  The joy stick moves and positions the cursor at a desirable object on the screen.

Source Data Automation
    The object of source data automation is to collect data without human intervention.  This process speeds up data collection, reduces errors, and gathers data one time and at the source of the transaction.
    Automated Teller Machines.  Automated teller machines (ATMs) are interactive input/output devices that enable people to obtain cash in many locations and updating their bank accounts instantly.  ATMs can handle a variety of banking transactions, including the transfer of funds to specified accounts.  One drawback of ATMs is their vulnerability to computer crimes and to attacks made on customers as they use outdoor ATMs.
    Point of sale terminals.  Many retail organizations utilize point of sale (POS) terminals.  The POS terminal has a specialized keyboard.  For example, the POS terminals at McDonald’s include all the items on the menu, often in labeled with the picture of the item.  POS terminals in a retail store is equipped with a scanner that reads the bar-coded sales tag.  POS devices increase the speed of data entry and reduce the chance of errors.  POS terminals may include many features such as scanner, printer, voice synthesis, and accounting software.
    Bar Code Scanners.  Bar code scanners scan the black-and-white bars written in a code called the Universal Product Code (UPC).  This code specifies the name of the product and its manufacturer.  Bar codes are especially valuable in high-volume tracking where keyboard entry is too slow and/or inaccurate.  Applications include supermarket checkout, airline baggage stickers, and Federal Express packages.  The wand reader is a special hand-held bar code reader that can read codes that are also readable by people.
    The Uniform Code Council (UCC) has proposed a 14-digit bar code called reduced space symbology (RSS) that, if adopted by the supermarket industry, allows produce to be tracked the same way other items in the store are.  Produce sections currently have fruits and vegetables with four digit bar codes.  The lack of digits prevents stores from differentiating certain produce items from others.  RSS will allow retailers to track international goods that are labeled using 13-digit codes and incorporate a 14th digit that represents a case of produce as one digit, allowing stores to sell in bulk more easily.
    Optical Mark Reader.  The optical mark reader is a special scanner for detecting the presence of pencil marks on a predetermined grid, such as multiple-choice answer sheets.
    Optical Character Reader (or Optical Scanner).  With an optical scanner, source documents such as reports, typed manuscripts, and books can be entered directly into a computer without the need for keying.  An optical scanner converts text and images on paper into digital form and stores the data on disk or other storage media.  Optical scanners are available in different sizes and for different types of applications.
    The publishing industry is a leading user of optical scanning equipment.  Publishers scan printed documents and convert them to electronic databases that can be referenced as needed.  Similarly, they may scan manuscripts instead of retyping them in preparation for the process that converts them into books or magazines.  Considerable time and money are saved, and the risk of introduction typographical errors is reduced.  Scanners are becoming increasingly more reliable.  Some are sophisticated enough to read not only text but visuals such as photos, illustrations, and graphs.
    Magnetic Ink Character Readers.  Magnetic ink character readers (MICR) read information printed on checks in magnetic ink.  This information identifies the bank and the account number.  On a canceled check, the amount is also readable after it is added in magnetic ink.
    Handwriting Recognizers.  Today’s scanners are good at “reading” typed or published material, but they are not very good at handwriting recognition.  Handwriting recognition is supported by technologies such as expert systems and neural computing and is available in some pen-based computers.
    Scanners that can interpret handwritten input are subject to considerable error.  To minimize mistakes, handwritten entries should follow very specific rules.  Some scanners will flag handwritten entries that they cannot interpret or will automatically display for verification all input that has been scanned.  Because handwritten entries are subject to misinterpretation and typed entries can be smudged, misaligned, and/or erased, optical scanners have an error rate much higher than the error rate for keyed data.
    Pen-based input devices utilize handwriting recognition.  These devices consist of a flat-screen display tablet and a pen-like stylus.  Users print directly onto the tablet-sized screen.  The screen is fitted with a transparent grid of fine wires that detects the presence of the specialized stylus, which emits an electronic signal from its tip.  The screen can also interpret tapping and flicking gestures made with the stylus.
    Pen-based input devices transform the letters and numbers written by users on the tablet into digital form, where they can be stored or processed and analyzed.  At present, pen-based devices cannot recognize free-hand writing very well, so users must print letters and numbers in block form.
    At high-tech meetings, it is not unusual to see everyone around a conference table using laptop computers.  In many settings, however, such behavior is unacceptable.  In Europe and Asia, it is completely unacceptable, and the use of laptops is prohibited in some places such as courtrooms and in the U.S. Congress.  Further, computer keyboards are not great for note-taking.  It’s difficult to type diagrams, flow charts, graphs, or equations into a computer.  As a result, A.T. Cross has developed an alternative interface to computers, called the CrossPad.
    The CrossPad is similar to an electronic clipboard.  The user slips in a pad, picks up the special pen supplied with the device, and can store up to 50 handwritten pages in the CrossPad’s memory.  Users can connect the clipboard to the serial port on a Windows 95 personal computer and a facsimile of the handwritten pages is input into the PC.
    Users could accomplish the same thing by running their written notes through a scanner, but the CrossPad is more convenient and comes with interesting IBM software.  If you tap a button on the CrossPad and then circle a word or phrase on your sheet of paper, it will be converted to text and recognized as a keyword when you copy the writing to your PC.  This process makes it simple to store indexed, searchable copies of handwritten documents on your computer.  Software also makes it possible to use the CrossPad to fill in printed forms, an approach that many health-care workers and others who spend much of their working lives filling in forms may find simpler and more natural than a laptop or handheld computer.
    Users must invest an average of 45 minutes in training the device to read their writing.  However, users find that they can fix any errors much faster than they could type in all the text.
    Voice Recognizers.  The most natural way to communicate with computers is by voice, using a natural language (called natural language processing).  Voice recognition devices convert spoken words into digital form.  Voice recognition devices are extremely important because they are fast, free the user’s hands, and result in few entry errors.  They also allow people with visual or other disabilities to communicate with computers.  When voice technology is used in combination with telephones, people can call their computers from almost any location.  While voice technologies have certain limitations such as the size of the vocabulary, they are rapidly improving.
    To understand a natural language inquiry, a computer must have sufficient knowledge to analyze the input in order to interpret it.  This knowledge includes linguistic knowledge about words, domain knowledge, common sense knowledge, and even knowledge about users and their goals.
    Sensors.  Sensors are devices that collect data directly from the environment for input into a computer system.
    Digitizers.  Digitizers are devices that convert drawings made with a pen on a sensitized surface to machine-readable input.  As drawings are made, the images are transferred to the computer.  This technology is based on changes in electrical charges that correspond to the drawings.  Digitizers are used by designers, engineers, and artists.
    Cameras.  Regular video cameras can be used to capture pictures that are digitized and stored in computers.  Special cameras are used to transfer pictures and images to storage on a CD-ROM.  A digital camera can take photos and load them directly from the camera, digitally, to a main storage or secondary storage device.
Hewlett-Packard produces the PhotoSmart C20 digital camera which sells for approximately $700.  Features include a 1.8 inch LCD screen, flash, camera-to-TV video port, autofocus, removable memory, and a lens cover that, when removed, powers up the camera.  The C20 has 1,152 x 872 pixel resolution.  Depending on the resolution or quality the user selects, the C20 stores beetween 8 and 40 pictures on its 4 MB compact flash pop-out memory card.

Output Devices
    The output generated by a computer can be transmitted to the user via several devices and media.  The presentation of information is extremely important in encouraging users to embrace computers.
    Monitors.  The data entered into a computer can be visible on the computer monitor, which is basically a video screen that displays both input and output.  Monitors come in different sizes, ranging from inches to feet, and in different colors.  The major benefit is the interactive nature of the device.  Monitors display information in a softcopy form, and they are used in most of the interface modes.
    Monitors employ the cathode ray tube (CRT) technology, in which an electronic “gun” shoots a beam of electrons to illuminate the pixels (picture elements) on the screen.  The more pixels on the screen, the higher the resolution.  For example, a screen with a 1024 x 768 resolution (786,432 pixels) would have greater clarity than one with a resolution of 640 x 350 pixels (224,000 pixels).  The dot pitch is the distance between pixels on the screen.  The smaller the dot pitch, the larger the number of pixels on a given screen and the higher the resolution.  A dot pitch of .28 mm or smaller is considered to be high quality.
    CRT monitors can be classified as monochrome or color and by their display capabilities.  Some monitors display only text, where others display both text and graphics.
    Portable computers use a flat screen consisting of a liquid crystal display (LCD).  Flat-panel displays are shallow-depth video displays based on solid-state technology.  Sizes vary from the 12-inch flat-panel screens found on notebook and laptop computers to the 17-inch flat-panel displays now being introduced for desktop computers, to wall-sized flat-panel monitors used in videoconferencing and home entertainment.
    LCDs consume less power than traditional monitors.  For example, LCDs consume less than half the power that a 15-inch desktop monitor uses.  However, today’s flat-panel displays are roughly six to eight times the cost of an equivalent CRT.
Passive-matrix LCDs are used in applications where display speed and color brightness are not essential.  More advanced active-matrix LCDs appear to function as one large integrated circuit, but actually are composed of hundreds of thousands of individually illuminated thin film transistors.  As a result, LCDs eliminate the flicker associated with CRTs.
    Currently, NEC Corporation is developing a 20-inch flat-panel display for desktop computers.  This panel increases the viewing angle to be comparable to that of standard monitors.  In addition, Sharp Electronics Corporation’s line of desktop replacement displays delivers viewing angles of more than 70 degrees in three directions.
    Screen sizes on notebooks and laptops are increasing.  Thirteen-inch screens are appearing on high-end laptops, the largest size practical for mass-production laptops because anything larger would be wider than the keyboard.
    Plasma displays are normally a few inches thick and weigh much less than CRT monitors.  As a result, plasma display technology is appropriate for large-screen displays such as those used on the trading floors of the New York Stock Exchange.
Field emission display (FED) is an emerging display technology that combines the best features of standard cathode-ray tube displays and flat-panel technologies.  Like CRTs, field emission displays rely on light emitted from phosphors that have been excited by electron beams.  However, like flat panels, FEDs are thin and can be made using standard semiconductor fabrication processes.  In addition, FEDs offer unlimited viewing angles.
    Impact Printers.  Like typewriters, impact printers use some form of striking action to press a carbon or fabric ribbon against paper to create a character.  Serial printers print one character at a time.  The most common serial printers are the dot matrix, daisy wheel, and line.  Line printers print one line at a time; therefore, they are faster than other serial printers.  Impact printers are slow and noisy, cannot do high-resolution graphics, are often subject to mechanical breakdowns,  but they are inexpensive.
    Nonimpact Printers.  Nonimpact printers overcome the deficiencies of impact printers, but they are more expensive.  Laser printers are higher speed, high-quality devices that use laser beams to write information on photosensitive drums, whole pages at a time; then the paper passes over the drum and picks up the image with toner.  Because they produce “print” quality text and graphics, laser printers are used in desktop publishing and in reproduction of artwork.  Thermal printers create whole characters on specially treated paper that responds to patterns of heat produced by the printer.  Ink-jet printers shoot tiny dots of ink onto paper.  Sometimes called bubble jet, they are relatively inexpensive and are especially suited for graphical applications when different colors of ink are required.
    Plotters.  Plotters are printing devices using computer-driven pens for creating high-quality black-and-white or color graphic images – charts, graphs, and drawings.  They are used in complex, low-volume situations.
    Voice Output.  Some devices provide output via voice – synthesized voice.  This term refers to the technology by which computers “speak.”  The synthesis of voice by computer differs from a simple playback of a prerecorded voice by either analog or digital means.  As the term “synthesis” implies, the sounds that make up words and phrases are constructed electronically from basic sound components and can be made to form any desired voice pattern.  The quality of synthesized voice is currently very good, and relatively inexpensive.

Multifunction Devices
    Multifunction devices combine a printer, fax machine, scanner, and copy machine into one device.  Multifunction devices are cheaper than buying specific devices separately and they take less space on a desktop.  The Hewlett-Packard LaserJet 3100 is a multifunction device.  This device prints at 600 dots-per-inch black-only output at six pages per minute.  It also adds scanning, copying, and a sophisticated fax machine to the package.
    In addition, the 3100 can be controlled through your personal computer using JetSuite software from JetFax.  Drop a document to be faxed or scanned into the input slot, and a menu pops up on your screen asking if you want to fax, copy, or scan the document.  For example, choosing “fax” allows you to fill in a cover page form, dial the phone, and send the document to one or more recipients.  You can create documents in a word processor or other application,, then dispatch them by “printing” to the fax machine.

MULTIMEDIA
 
    Multimedia integrates, using information technology, any or all of the following:  text, sound, still images, animation, and digitized motion video.
            text:  narrative descriptions, including words, descriptions, and symbols;
            still images:  photographs, drawings, and figures, whether in black and white, shades of gray, or full color;
            audio sequences:  sounds of human voices, music, special effects;
            motion video sequences:  integration of sound and full motion pictures;
            animation sequences:  succession of discrete still images that, when played under computer control, are interpreted by the human mind as a smooth-flowing moving picture.
    Multimedia merges the capabilities of computers with televisions, VCRs, CD players, and other entertainment devices.
Multimedia invites participation by the receiver of the information, typically by way of a graphical user interface.  The most important advance in multimedia arising from information technology is the capability for dynamic, interactive presentation of information.  Today, recipients of multimedia presentations have the option of making the communication two-way.
    An interactive multimedia approach involves the use of computers to improve human-machine communication by using a combination of media.  The construction of a multimedia application is called authoring.
    Multimedia standards are necessary to enable software and hardware vendors to build products that will work together to meet customer requirements.  The Multimedia Personal Computer Council (MPC) was formed by Microsoft and hardware manufacturers such as NCR, Fujitsu, and NEC.  The MPC specification defines the minimum hardware requirement for multimedia presentations.  The MPC certification mark tells consumers that the hardware has been tested by the manufacturer and found to be compliant with the minimum hardware specifications.
    IBM has provided the Ultimedia Solution, which employs Intel digital video interactive technology and Phillips CD-I, compact disk interactive technology.  This standard defines the operating system, hardware, and software for interactive compact disk applications.  Compact disks are the delivery medium of choice for multimedia applications, because CDs have storage capacities of 650 megabytes.  Video, image, and audio compression technologies make it possible to deliver multimedia applications that are rich in content and features while using less than 600 megabytes of storage.
    Audio involves converting an analog signal to a digital recording for storage on a magnetic disk or CD-ROM, and then converting the digital recording to an analog system when the multimedia program executes.  Audio devices include CD-audio and cassette players.
    The standard system for connecting musical instruments and synthesizers to computer is the Musical Instrument Digital Interface (MIDI).  MIDI defines codes for musical events, including the start of a note and its pitch, length, volume, and other attributes.  A MIDI card can be added to a personal computer to control audio input and output.  Users can edit music by changing the various codes in the digital recording.
    Advanced sound systems use digital signal processor (DSP) chips to improve the analog-to-digital-to-analog conversion process.  DSP chips perform signal conversion instead of the personal computer’s CPU.
    Video, a rewarding part of multimedia presentations, is also the most difficult element to display because a single uncompressed frame requires about 1 megabyte of storage.  It is possible to avoid this limitation through video compression, where only the parts of successive frames that change are stored.
    Intel has developed special-purpose chips for compressing and processing full-motion video.  The Intel standard is called digital video interactive (DVI).  DVI compresses video at a 150:1 ratio, enabling one hour of video to be stored on about 720 megabytes.  Without compression, over 110 gigabytes would be required.
 

                             Total Cost of Ownership

Business Problem
    Norwest Corporation is trying to save $90 million per year.  The company feels that if it can decrease the costs of purchasing and maintaining the firm’s 30,000 personal computers by 30%, it can reach its goal.  The question is, how will Norwest do it?
    The price tag on a computer has become an increasingly irrelevant indicator of what it really costs.  One of the most important issues facing organizations today is the total cost of ownership for their computers.  In spite of rapidly decreasing hardware costs, the total cost of ownership continues to increase.
    The initial purchase of computing equipment is relatively small when compared with the total cost associated with maintaining and supporting the entire information technology process in organizations today.  Other activities such as user support, training, networking, and other facilities-related operations contribute the majority of cost of ownership of an enterprise information system.
    Equipment costs are fairly straightforward.  The acquisition price and ongoing maintenance cost are relatively clear and are usually the starting point for a full cost-of-ownership analysis.  However, there are many other factors that must be considered.  To determine a complete cost of ownership, the following types of questions must be answered:
    What are the future upgrade costs?
    What type of reliability and high availability capabilities are required?
    What level of user support will be required?
    Where are the users located?
    What level of training is needed?
    Are the users computer-literate?  (e.g., engineers or clerical staff)
    How will remote users be supported?
    What level of security is needed?
The Gartner Group has developed a model for total cost of ownership that includes the costs of four areas:  end-user operations, capital, administrative, and technical.  End-user operations (40% of the TCO) costs consist of data management, learning time, peer support, and small, user-solvable problems.  Capital costs (20% of TCO) include hardware, software, and options.  Administrative costs (20% of TCO) cover asset management, security, auditing, installation and change management.  Technical-support costs (10% of TCO) include the help desk, configuration review, and data extraction.  On an annual basis, the Gartner Group estimates that the TCO for each networked personal computer is approximately $11,900.

The Business Solution(s)
    In an effort to reduce their TCO and gain flexibility, many organizations are choosing to lease their computers.  These firms say that leasing allows them to keep up with rapidly changing technology and leaves them more cash to expand their businesses.  In addition, the cost of a lease can be deducted and a leased computer can be simply shipped back.  This process saves companies the effort and time spent in finding a buyer for older computers.  Popular lease plans run for three years with the option to buy the machine at the end of the lease.  Hollywood Entertainment Inc., in a pilot study using network computers in five of their video stores, found that maintenance and support costs were cut in half.
    There are several other methods that companies use to reduce their TCO.  First, standardized environments cost less to install and maintain than heterogeneous environments.  For example, Lexmark International, a printer manufacturer, will not develop applications, but relies on buying standard, industry-leading products from major software vendors.  Lexmark’s consistent desktop environment reduces training costs and reduces time lost from employees tinkering with their computers.  Lexmark employs 80 people to support their 7,500 personal computers worldwide.
    Electronic software distribution ensures consistent software installation and eliminates the need to physically install software on each computer.  People’s Bank, a $7.9 billion financial institution based in Bridgeport, Connecticut, uses electronic software distribution and has reduced its TCO to just under $5,000 per year per computer.
    The use of remote systems management tools to move software and data to and from personal computers and to store backup images of users’ hard drives will also reduce TCO.  Atlanta-based HBO & Co., a large healthcare services provider, has more than 800 mobile sales representatives, all using laptop computers.  With a standardized disk image on each machine and a standardized operating environment, HBO’s representatives dial in from hotels, homes, and client offices all over the United States, uploading sales reports and downloading the latest competitive intelligence and product information.  Any settings that users have changed are changed back automatically and any unauthorized software is deleted automatically.  Interestingly, each user is allowed a 50 MB data area for ‘personal control.’  HBO uses archived disk images from each laptop to handle laptop damage, theft, and virus attacks.
    Organizations may employ automated technical support tools to reduce support staff.  The Gartner Group estimates that an organization with 1,000 network nodes averages 72 help desk calls per day, with each call costing between $22 and $41 to resolve.  The National Institute of Allergy and Infectious Diseases at the National Institutes of Health uses automated support tools to allow each support person to handle 100 to 125 users.  The automated support tools also show support personnel what users have done to bring about the problem that prompted the call to the help desk.
    What did Norwest do?  The financial services company decided to purchase NetPCs.  NetPCs, slimmed-down versions of regular, desktop personal computers which cost approximately $1000, fall in between regular PCs and network computers.  Where network computers are $750 computers with no hard drive, NetPCs have a hard drive and can have programs installed and run locally.  Norwest also has a standard, corporate-wide computing environment, distributes software electronically over its network, and also performs system administration tasks over the network.
    Question:  Rank the methods for reducing TCO from the greatest effect to the least effect.  Discuss your ranking.