D.

Disk System Architecture

The simple fact of life is that all information needs to be stored somewhere to make them easily
accessible and retrievable. This storage space needs to be large enough to hold as much information
as possible, but it should also be organized for easy access to that information. Although some
people simply use the top of the desk and store documents in piles, the best solution for storing
paper documents at most offices is a filing cabinet.
With computers, you have various types of electronic information to store, including data files,
application files, and configuration files. Under this chapter, we will explain the following storage:
Storage types, Floppy disk systems, IDE disk systems and Small Computer Systems Interface (SCSI)

Storage Types
This section gives a detailed description of each type of disk storage, starting with a brief overview of
punch card and tape storage and moving into the different types of disks.

1. Punch Cards and Tape

Originally, computer information was contained in memory and printed to punch cards. When you
wanted to retrieve the data, you ran the punch cards back into a special reader; the information was
read back into memory and could then be used. This proved to be a very limited storage medium in
terms of capacity and ease of retrieval. A single page of data could take several pages of punch
cards. Also, the punch card order was very important. The cards had to be put into the reader in the
correct order or the program wouldn’t run correctly. Hopefully you kept the cards in their correct
order - God forbid you dropped them on the floor or some jokester shuffled them.

Then one day, someone discovered that computer signals could be recorded with a tape recorder.
When the tape was played back into the computer, the information was retrieved. This was a much
more efficient storage system than punch cards. Because the tape moved in only one direction and
was a single, long tape (rather than several punch cards), it reduced the possibility of getting the
information out of sequence and made it easier to access the information.

We’ve all seen movies where the computer room has massive cabinets with reels of tape moving
back and forth. What you might not know is that several early personal computer systems (such as
the TRS-80, the Tandy Colour Computer, and the Apple II) came with cassette tape storage devices to
load and store programs and data.

The major limitations of tape were in speed and precision. Tape was slow to store programs and
data. Also, it was slow to access information because of its linear nature. Tape devices are described
as sequential storage devices. With these types of devices, if a piece of information is located at the
end of a tape, you have to “fast forward” through all the other information to get to the data you
need. Also, the early tape mechanisms didn’t have very good position locators or stepper motors. It
was quite possible to start reading a tape after the information you needed had already begun,
causing you to miss the start of the program or data and making it unusable.

Despite its drawbacks for most tasks you might want to undertake with a computer, magnetic tape
in various forms is still a great medium for “backing up” your system - that is, recording some or all

1
of the information stored on your hard disk and putting the recording away in case of hard disk
catastrophe.

Why is tape, which is not recommended for other types of information storage, recommended so
often for backups? In part, it’s because of the sheer amount of data it can store. Some of the cheaply
available types of tape media of today can hold more than 4GB and are easily movable, removable,
and insertable into their respective recorder/playback drives. Moreover, the relatively low speed of
recording or playing back a backup tape is not a significant consideration for most users as it would
be with other computer tasks. Most users know that they are simply performing a safety measure
when they utilize a backup, so they schedule their backups to be carried out automatically when no
one is using the computer anyway.

2. Disk Drives
To overcome the limitations of magnetic tape, magnetic disk systems were developed. Rotating
stacks of disks were coated with a special substance that was sensitive to magnetism. As the disk
rotated, the particles in the substance could be polarized (magnetized), indicating a 1. The un-
polarized areas would indicate 0s.
This technology is the cornerstone of most current disk storage types and hasn’t changed much since
its inception. These types of disk systems store data in nonlinear format and are called random
access storage devices, in contrast to tape’s sequential access storage methods. The data can be
accessed no matter where it is located on the disk because the read/write head can be positioned
exactly over the requested data so you don’t have to “fast forward” through the data that was
stored before it.

There are two major types of disk systems - fixed and removable.

Fixed Disk Drives - Also known as hard disks and hard drives, fixed disks actually contain several
disks called platters, stacked together and mounted through their centres on a small rod called a
spindle (see Figure 4.1). The disks are rotated about this rod at a speed between 2,000 and 10,000
revolutions per minute (RPM). As the disks rotate, one or more read/write heads float approximately
10 microinches (about 1/10th the width of a human hair) above the disk surfaces and make, modify,
or sense changes in the magnetic positions of the coatings on the disks. Several heads are moved
together as one unit by an actuator arm. There is usually one head for each side of a platter. This
entire mechanism is enclosed in a hard disk case. These disks are called fixed disks because the
mechanism is not designed to be removed. The disk platters, though perfectly free to revolve at high
RPM, are otherwise fixed in place.

 Disk Organization - We must have a way of organizing this disk into usable sections. It’s done
by first dividing up the platters into sections as you would a pie and then further dividing this
area into concentric circles, called tracks (see Figure 4.2). Tracks are numbered from the
outside (track 0) to the inside (track 902 on a 903-track hard disk). A disk sector is the part of a
track that falls in a particular section of the “pie slice” on the disk.

If you can visualize several tracks stacked together vertically (the same tracks on each disk),
you might describe that collection of tracks as a cylinder. In fact, that’s how tracks are referred

2
 to in a discussion of the organization of disk information, because when information is read
from or written to a disk, the heads read or write a sector-sized division of a track a whole
stack at a time, from top to bottom (see Figure 4.3). In other words, the disks’ tracks aren’t
treated as individual tracks on single disks; they’re treated as cylinders. This amounts to quite
a bit of information read or written at one time. The precise amount of information that is
read at once depends on the number of cylinders, heads, and sectors, or what we call drive
geometry.

 Disk Specifications - When a fixed disk is rated, it has several qualities that are given as
specifications.
Size, or capacity, is just one of the properties given. Size is determined by drive geometry. Let’s
quantify this by giving some values to each variable. Let’s say we have 903 cylinders, 12 heads
(6 platters—1 head per side, 2 sides to each platter), band 63 sectors per track. A typical fixed
disk has 512 bytes per sector (that’s 0.5KB/sector). The capacity of the hard disk would be
341,334KB, or 333MB (903 cyl × 12 heads × 63 sectors/track × 0.5KB/sector). These values are
commonly given on the outside of a fixed disk.

Another quality that is used to gauge the performance of an individual drive is seek time. This
value, commonly given in milliseconds (ms), is how long it takes the actuator arm to move
from rest to the position where the read/write head will access information. Additionally,
because the platters rotate, once the read/write is in position, it may take a few milliseconds
for the target sector to move under the read/write head. This delay is known as the latency
factor. Latency values are given in milliseconds (ms).

Because the faster a drive spins, the lower the average latency values, drives are also rated
with a spin speed. Spin speeds indicate how fast the platters are spinning. They are stated in
revolutions per minute (RPM); higher RPM values mean faster speeds and lower latency
values.
For example, if a disk has a spin speed of 5,000 RPM, it will be rotating at 83.34 revolutions per
second. One rotation will take 1/83.34 seconds, or 11.9ms. This value would represent the
largest possible delay (i.e., the disk head is in position just after the required sector moved
past it). However, disk latency values are actually only average latency values, because latency
values are not constant. The target sector could be close, or it could be far away. So in our
example, we take the largest possible latency (11.9 ms) and smallest possible latency (0 ms)
and average them to get the average latency: (11.9 ms + 0)/2 = 5.95 ms.

The two factors, seek time and latency, make up the drive’s access time. Put simply, the
average seek time plus the average latency equals the drive’s access time. The smaller the
access time value is, the faster the drive.

 Read/Write Processes - Now that we know how the drive functions and how the drives are
rated, how the drive stores the information. At the most basic level, the disk works by making
flux transitions with an electromagnet in the read/write head to store information on the disk
surface.

3
 A flux transition is the presence or absence of a magnetic field in a particle of the coating on
the disk. As the disk passes over an area, the electromagnet is energized to cause the material
to be magnetized in a small area. The process by which binary information is changed into flux
transition patterns is called encoding.

There are many ways of converting 1s and 0s to flux transitions. The simplest way is to
interpret the presence of a flux transition as a 1 and the absence as a 0. Because this was the
most obvious choice, the first hard disks (ST-506, ESDI types) used this method of encoding,
known as Frequency Modulation (FM), and its cousin Modified FM (MFM). This worked well
until techniques for increasing the track/cylinder density became almost too successful. What
happened was that tracks would be placed so tightly together that at higher speeds, the
read/write heads would affect not only the track immediately under the head, but the
adjacent ones as well.

To solve this problem, a technology known as Run Length Limited (RLL) was developed; it
spaced the 1s farther apart using a special code for each byte. This method turned out to be
more efficient for large drives than for small ones. RLL encoding also introduced data
compression, a set of technologies that increased the amount of data that could be stored on
the drive. Most of today’s drives (IDE, SCSI, and so on) use a form of RLL encoding.

With this new type of encoding, much more data could be transferred to the computer at
once, but this created a new problem. The interface would sometimes get bogged down and
would stop reading in order to “catch up.” This was a problem because during this pause, the
platters were still rotating and the read/write heads could skip a whole bunch of sectors. To
solve this problem, disk designers developed a technology known as interleaving.

To make it easier for the operating system to manage the storage space, the information
encoded on the drive is written to groups of sectors known as clusters. A cluster is made up of
up to 64 sectors grouped together (the actual number of sectors included in a cluster varies
with the size of the hard disk). When the operating system is storing information, it writes it to
a particular cluster instead of to an actual sector because it’s more efficient for the operating
system to keep track of clusters than sectors. The file that contains the information about
where the tracks and sectors on the disk are located is known as the file allocation table, or
FAT. It is contained in the outermost track (track 0) of the disk.

 Formatting the Disks to Prepare Them for Use - To create the FAT, a machine at the factory
performs a procedure known as a low-level format. This procedure organizes the disk into
sectors and tracks. Once the sectors and tracks have been created, the low-level format
procedure makes the FAT file and records in it the positions of the new sectors and tracks.
Additionally, during this procedure, the low-level format procedure meticulously checks the
disk’s surface for defects. If any are found, the locations of these “bad spots” are entered into
the FAT as well, so that the operating system knows not to store any information in those
locations.
When an operating system is installed, it will do a high-level format (or operating system
format) and create its own separate FAT that keeps track of where clusters are located and

4
 which files are located in which clusters. These two FAT technologies are applicable to all types
of drives.

Removable Media Drives - Another type of drive system is removable media drives. Removable
drives use technologies similar to fixed disk (fixed media) drives, except the storage medium is
removable. The obvious advantage is that removable media multiplies the usefulness of the drive.
With a hard disk, when the disk is full, the only two things you can do to increase space are to delete
some information or get a larger drive. With removable media drives, you can remove the full disk
and insert a blank one. Other than the Zip and Jaz drives, there are numerous categories of
removable media drives covered under this topic, as discussed.
 Floppy Disk Drives - The type of removable media drive that is the most often described is the
floppy disk drive. It is called that because the original medium was flexible. Floppy disks are
like fixed disks, but they have only one “platter” encased in a plastic shell.

There are several types of floppy disk drives available in many different capacities (the figure
below shows a 1.44MB 31⁄2-inch drive).

A floppy disk drive

A floppy drive has either one or two read/write heads. Each head moves in a straight line on a
track over the disk rather than on an angular path as with fixed disk systems. When the disk is
placed into the drive, a motor engages the centre of the disk and rotates it. This action moves
the tracks past the read/write heads.
 CD-ROM Drives - Another type of removable media drive is the CD-ROM drive. These drives
are slightly different from other storage media in several ways. First of all, they have a
different way of reading information than magnetic media disk drives do. Because CD-ROM
drives use laser light to read the information from the media, they are described as optical
drives.

5
A CD-ROM drive

When reading information from a CD, the drive is basically reading a lot of pits and lands
(lands are the spaces between the pits) in the disc surface. The pits are etched into the CD at
production time. The laser reflects off the CD’s surface and onto a sensor. The sensor detects
the pattern of pits and lands as the disc rotates and translates them into patterns of 1s and 0s.
This binary information is fed to the computer that is retrieving the data.

Another difference between magnetic media and CD-ROM drives is that CDROM drives are
read-only devices (CD-ROM stands for Compact Disc–Read-Only Memory). The only way of
writing to a CD-ROM is during manufacture time, where the pits are “burned” into the
substrate of the disc. Once written, they cannot be erased.

The reason CD-ROMs are so popular, even though they can’t be written to, is their large
capacity (greater than 500MB) and easy access. They are a great choice for archival storage.
Most often used for software distribution, they have really taken off in the past few years as
the speed of CD-ROM drives has increased.

Finally, a CD-ROM disc has a single track that runs from the centre to the outside edge, exactly
the reverse of the groove on a record. A CD-ROM uses basically the same technology as the
audio compact discs in use in most homes today. When a CD-ROM is placed into a CD-ROM
drive, a motor spins the CD at a specific rate. A laser that reads the CD is then activated.
Because of these basic similarities, there are several compatibilities between the different
compact disc technologies. For example, it is possible to play audio CDs in a computer’s CD-
ROM drive. Also, some computer CDs have audio tracks on them and are made to be used in
either type of CD drive (home audio or computer).

 Optical Disk Drives - Another type of disk is the optical disk. An optical disk is much like a CD-
ROM except that optical disks can be read from and written to (like fixed disks, except optical
disks are similar to CD-ROM discs and thus a laser is used). The upside is that optical disks are
removable and can hold lots of information. The downside is that they are expensive and are
still slower than fixed disks.

6
  Tape Drives - The final type of removable media drive is a tape drive (see Figure 4.6). The
tape cartridge uses a long polyester ribbon coated with magnetic oxide wrapped around two
spools. As the tape unwinds from one spool, it passes by a read/write head in the drive that
retrieves or saves the information. It then proceeds to the other spool where it is kept until
needed again.

A tape drive
Tape media is great for large-capacity storage, but it is agonizingly slow. The best application
for tape media is for backup purposes. Current tape technology uses 4mm or 8mm Digital
Audio Tape (DAT) or Digital Linear Tape (DLT) for its storage medium. With these technologies,
it is possible to store up to 70GB of data on a single tape cartridge.

Floppy Disk Systems

There are four major types of floppy disk drives in use today. They are usually identified by the size
of the media they use and the capacity of that media. It important to note that there are four
components of a floppy drive subsystem: the disk (also called the medium), the drive, the controller,
and the cable.
 The Medium - The floppy disk, as we have already mentioned, is the removable medium on
which information is stored in a floppy disk system. There are two major types in use today,
the 51⁄4-inch “floppy” disk and the 31⁄2-inch “diskette” (sometimes incorrectly called a “hard
floppy” or even a “hard disk”).

A 51⁄4-inch floppy disk and a 31⁄2-inch diskette

The 51⁄4-inch floppy disks are made from a polyester disk coated with iron oxide and a flexible
outer covering. The disk has a large hole in the centre, called the drive hole that is used by the
motor in the disk drive to spin the disk. In addition, there is also a 11⁄2-inch oval window cut
into the case to allow the read/write heads access to the disk media. A small round hole cut
into the disk shell next to the drive whole lines up with an even smaller hole cut into the disk
media. When this smaller hole spins past the slightly larger hole in the shell, it allows a light to
shine all the way through the disk system. In this way, the floppy drive can tell how fast the
disk is rotating by how many times in a second that hole appears.
Finally, there is a notch cut in one side of the disk. This notch is called the write-protect tab.
When a disk is inserted into a floppy drive, a small lever places itself into this notch. When the
lever is in the notch, the disk can be written to. You can “write protect” the disk (which

7
 prevents it from being written to) by covering this hole. The 51⁄4-inch disks have almost
completely disappeared in the last few years, replaced by the 31⁄2-inch format, and most newer
computers no longer include 51⁄4-inch drives.

The other type of floppy disk media is not really “floppy” at all. Some people mistakenly call it
a “hard disk.” Its real name is a 31⁄2-inch diskette (to differentiate it from a full-grown “disk,”
we suppose). The 31⁄2-inch diskettes are also made from a polyester disk coated with a layer of
iron oxide. This disk is enclosed in a durable, plastic case. This was an improvement over the
51⁄4-inch variety because the 51⁄4-inch floppies were easily creased or damaged. Also, the 31⁄2-
inch diskettes have a metal shutter over the media access window. Again, this was an
improvement over 51⁄4-inch media because people often grasped the disks inadvertently by this
edge of the disk, pressing their fingers onto the media and thus contaminating the disk,
making it difficult to read.
Finally, there is a notch with a sliding plastic tab over it to write protect the 31⁄2-inch disk. This
is also better than 51⁄4-inch disks because the write-protect notch in the other disks was
covered with a type of tape that could come loose in the drive and cause gum-ups.

 The Drives - The next item we need to discuss is the floppy disk drives themselves. There are
three items we need to discuss regarding the drives: media size, form factor, and capacity.
Media Size - Because we just covered media size in our discussion of the media, we’ll just refer
you to the previous paragraphs and continue on. We do want to make one note, however. You
can’t put a 51⁄4-inch diskette into a 31⁄2-inch drive. Or vice versa.
Form Factors - There are three main types of form factors, or drive styles, available today. A
form factor usually just means the physical dimensions and characteristics. Today’s floppy disk
drives use either the full-height, half-height, or combo form factors. Full-height drives were
the only ones available for the first PCs. The drives were large and bulky and usually lower in
capacity than today’s drives. Half-height drives take up only half the space (vertically) of full-
height drives.
The final form factor is becoming more popular as space becomes a premium in computer
cases. The combination form factor (or combo, for short) contains both 1.2MB 51⁄4-inch and
1.44MB 31⁄2-inch drives in a half-height enclosure. This has the obvious advantage of having
both drives in the space for one.
Capacity - The final topic in our discussion of drives is their capacity. The capacities range from
360KB to 1.44MB in various form factors. The Table below gives detail range of capacities
available today, their associated form factors, the number of sides used, and the density of the
disk. Some disks use only one side of the media, whereas others use both sides. The density of
a disk determines how closely the sectors and tracks can be packed. Notice how the
combination of all three items relates to the capacity.
Capacity Form Factor Density Sides
180KB 51⁄4" FH Single Double
360KB 51⁄4" FH Double Double
720KB 31⁄2" HH Double Double
1.2MB 51⁄4" FH/HH High Double
1.44MB 31⁄2" HH High Double
Table 1. Disk Capacities

8
  The Controller -The floppy disk controller is the circuit board that is installed in a computer and
translates signals from the CPU into signals that the floppy disk drive can understand. Often, the
floppy controller is integrated into the same circuit board that houses the hard disk controller.
Or even better, the controller might be integrated into the motherboard in the PC.

A floppy cable

 The Cable - The last topic on our list of prerequisites is the floppy drive cable. This cable is
made up of a 34-wire ribbon cable and three connectors.
One of these connectors attaches to the controller. The other two connect to the drives (one
for drive A:, the other for drive B:). You can attach up to two floppy drives to a single
controller. These connectors are specially made so that they can be attached to the drives in
only one way.

Additionally, you might think that with 34 wires, the data would be transmitted in a parallel
manner (8 bits at a time), but it isn’t. Data is transmitted 1 bit at a time, serially. This makes
floppy transfers slowly, but usually the serial data transfer isn’t the bottleneck.
The cable also has a red stripe running down one side. This stripe indicates which wire is for
pin #1 on the controller and on the drive(s). This pin is usually marked with a small 1 or a white
dot on the controller. When connecting the drives to the controller, you need to make sure
that the red stripe is oriented so the wire it represents is connected to pin #1 on the drive and
pin #1 on the controller. Also, you connect the drive that is going to be drive A: after the twist
in the floppy cable for most ISA floppy systems (very important).

Installation and Configuration

The first step is to connect the cable to the drive as per the aforementioned instructions. Then,
install the drive in the computer by physically mounting it in the computer case and connecting the
power cable. Finally, you connect the cable from the drive to the controller, as we discussed already.
With regular diskette drives, when connecting the cable to the drives, you connect the B: drive to the
connector before the twist in the cable and the A: drive to the connector after the twist.

Troubleshooting the floppy drive is also relatively simple. The most common problem after
installation is a drive light that refuses to go out. This is caused by having the floppy drive cable
upside down on one side. Most floppy cables are keyed so that they go on in only one direction.
However, in some systems, the floppy cable might not be keyed, so you must understand the
consequences of not having the cable on in the right direction. Just remember which way the red
stripe goes!

9
In addition to the cable position, you must also set the drive type and size of the floppy drive in the
CMOS setup. This is done so that the computer will know what drives are attached. If this
information is wrong, most computers will detect that the wrong drive type is selected and an error
will occur during boot-up.

You can also troubleshoot sporadic read/write problems. More often than not, these are caused by a
dirty drive. If this is the problem, it can be fixed with a floppy-disk head-cleaning kit. Another cause
might be a bad floppy disk. Floppies have a finite number of uses in them and they can go bad. (This
may come as a shock to some people who think their data is safe forever on a floppy disk.)
This problem can be fixed by copying the data (if possible) from the old disk to a new one.
Also, don’t forget to check the obvious things like disconnected floppy cable, power not plugged in,
or disks not inserted properly. Any one of these can cause problems that most people just assume
can’t be the problem because it’s too obvious.

IDE Disk Systems

ST-506 and ESDI drives were great drive systems in their day, but by the time the IBM AT computer
came out, drive sizes were increasing beyond the capabilities of existing technologies to utilize them.
The only way out was to use a technology called SCSI (pronounced “scuzzy,” it stands for Small
Computer System Interface. Although it was technologically superior, it was also, at the time, quite
expensive. Two companies, Compaq and Western Digital, saw this problem and developed an
alternative to SCSI.
Their idea was that if they could develop a cheap, flexible drive system, people would buy it. They
were more right than they realized. That solution, known as Integrated Drive Electronics (IDE)
because its major feature was a controller located right on the disk, is one of the most popular drive
interfaces on the market today.
IDE Technologies - The idea for IDE (more commonly known as AT Attachment interface, or ATA)
was a simple one: Put the controller right on the drive itself and use a relatively short cable to
connect the drive/controller to the system. This had the benefits of decreasing signal loss (thus
increasing reliability) and making the drive easier to install.

In addition, because the controller was integrated into the same assembly as the drive, the only
board that needed to be installed in the computer was an adapter that converted signals between
the motherboard and the drive/controller. The board is normally called a pass-through or paddle
board. (This board is often, incorrectly, called a controller. The term is incorrect because the paddle
board is often integrated with a floppy controller, two serial ports, a game port, and a parallel port.
In fact, this combination is normally called a multifunction interface board.) With some of today’s
systems, the IDE adapter is integrated into the motherboard.

Installation and Configuration - Installation and configuration of IDE and EIDE devices is much easier
than it is with ST-506 and ESDI systems.
The basic steps for installing them are the same: Mount the drive in the carrier, connect the cable to
the drive, install the drive in the computer, and configure the drive. However, IDE’s cabling and
configuration issues are less complex. For example, you only have a single, 40-pin cable to connect
the drives to the computer. (And no, there are no any twists in this cable.) Cabling is just a matter of

10
connecting the drive(s) to the cable and plugging the cable into the paddle board. This is made easier
because the majority of the IDE cables today are keyed so that they only plug in one way.

Small Computer Systems Interface (SCSI)

The last type of disk subsystem is probably the most flexible and the most robust. Conversely, it’s
probably also the most complex.
SCSI was a technology developed in the early ’80s and standardized by the American National
Standards Institute (ANSI) in 1986. The standard, which is known as ANSI document X3.131-1986,
specifies a universal, parallel, system-level interface for connecting up to eight devices (including the
controller) on a single, shared cable (called the SCSI bus). One of the many benefits of SCSI is that it
is a very fast, flexible interface. You can buy a SCSI disk and install it in a Mac, a PC, a Sun
workstation, or a Whizbang 2000, assuming they make a SCSI host adapter for it. SCSI systems are
also known for their speed. At its introduction, SCSI supported a throughput of 5MB/sec (5 to 10
times faster than previous buses).

SCSI devices can be either internal or external to the computer. If they are internal, they use a 50-pin
ribbon cable (similar to the 40-pin IDE drive cable). If the devices are external, they use a thick,
shielded cable with Centronics-50 or male DB-25 connectors on it. These devices aren’t always disk
drives. Scanners and some printers also use SCSI because it has a very high data throughput.

To configure SCSI, you must assign a unique device number (often called a SCSI address) to each
device on the SCSI bus (also sometimes call the SCSI chain). These numbers are configured through
either jumpers or DIP switches. When the computer needs to send data to the device, it sends a
signal on the wire “addressed” to that number. A device called a terminator (technically a
terminating resistor pack) must be installed at both ends of the bus to keep the signals “on the bus.”
The device then responds with a signal that contains the device number that sent the information
and the data itself.

This information is sent back to the SCSI adapter, which operates somewhat like a controller and
somewhat like a paddle board. The adapter is used to manage all the devices on the bus as well as to
send and retrieve data from the devices.

The adapter doesn’t have to do as much work as a true controller because the SCSI devices are
“smart” devices; they contain a circuit board that can control the read/write movement. It can also
receive signals like “Get this information and give it to me.” When a device receives a command like
that, it is smart enough to interpret the signal and return the correct information.

Types of SCSI
The original implementation of SCSI was just called “SCSI” at its inception. However as new
implementations came out, the original was referred to as “SCSI-1.” This implementation is
characterized by its 5MBps transfer rate, its Centronics-50 or DB-25 female connectors, and its 8-bit
bus width. SCSI-1 also had some problems. Some devices wouldn’t operate correctly when they
were on the same SCSI bus as other devices. The problem here was mainly that the ANSI
SCSI standard was so new that vendors chose to implement it differently. These differences would
be the primary source of conflicts.

11
After a time, the first major improvement to SCSI-1 was introduced. Known as SCSI-2 (ANSI Standard
document X3.131-1994), it improved SCSI-1 by allowing for more options. These options produced
several subsets of SCSI-2, each having its own name and characteristics. But the most obvious
change from SCSI-1 is that SCSI-2 now uses a higher-density connector. Also, SCSI-2 is backward
compatible with SCSI-1 devices.

The first improvement that was designed into SCSI-2 was a wider bus. The new specification
specified both 8-bit and 16-bit buses. The larger of the two specifications is known as Wide SCSI-2
because it’s wider (Captain Obvious rides again). It improved data throughput for large data
transfers. Another important change was to improve upon the now-limiting 5MBps transfer rate.
The Fast SCSI-2 specification allowed for a 10MBps transfer rate, thus allowing transfers twice as fast
as SCSI-1. So, Wide SCSI-2 transfers data 16 bits at a time, Fast SCSI transfers data 8 bits at a time,
but twice as fast (at 10MBps).

Another option was to combine both into a blazingly fast technology known as SCSI-2 Fast-Wide. It
combined the speed of Fast SCSI-2 with the bus width of Wide SCSI-2 to produce a transfer rate of
40MBps.
Finally, there is a new SCSI standard, SCSI-3 One of the feature sets is known as Fast-20 SCSI (also
known to some as Ultra SCSI). Basically, this is a faster version of Fast SCSI-2 operating at 20MBps for
narrow SCSI and 40MBps for Wide SCSI. Another feature set is the Ultra2 Low Voltage Differential
(LVD), which increases the maximum SCSI bus length to 25 meters (82 feet) and increases the
maximum possible throughput to 160MBps (on Ultra2 Wide LVD).

There are other proposed SCSI implementations, like Apple’s FireWire, Fibre Channel, and IBM’s SSA,
all offering speeds in the hundreds of megabit per second (MBps) range. Fibre Channel, specifically,
is gaining support in the LAN arena because of its high-speed storage access and shared-media
capability. The name Fibre Channel is somewhat of a misnomer because the technology will run over
fibre optic cable, STP, or coaxial cable.

SCSI Device Installation and Configuration

Installing SCSI devices is rather complex, but you still follow the same basic steps as mentioned with
the other types of drives (refer back to the previous types of hard disk interfaces if you’re still
unclear). The main issues with installing SCSI devices are cabling, termination, and addressing.
We will consider termination and cabling together because they are very closely tied together. There
are two types of cabling:
 Internal cabling uses a 50-wire ribbon cable with several keyed connectors on them. These
connectors are attached to the devices in the computer (the order is unimportant), with one
connector connecting to the adapter.
 External cabling uses thick, shielded cables run from adapter to device to device in a fashion
known as daisy-chaining. Each device has two ports on it (most of the time). When hooking up
external SCSI devices, you run a cable from the adapter to the first device. Then you run a
cable from the first device to the second device, from the second to the third, and so on.

12
Because there are two types of cabling devices, you have three ways of connecting them. The
methods differ by where the devices are located and whether or not the adapter has the terminator
installed. The guide to remember here is that both ends of the bus must be terminated:
Internal devices only - The first situation we’ll discuss is one in which you have internal devices only.
When you have only internal SCSI devices, you connect the cable to the adapter and to every SCSI
device in the computer. You then install the terminating resistors on the adapter and on the last
drive in the chain only. All other terminating resistors are removed.
External devices only - In the next situation, you have external devices only. By external devices, we
mean each with its own power supply. You connect the devices in the same manner you connected
internal devices, but in this method you use several very short (less than 0.5 meters) “stub” cables to
run between the devices in a daisy chain (rather than one, long cable with several connectors). The
effect is the same. The adapter and the last device in the chain (the one with only one stub cable
attached to it) must be terminated.
Both internal and external devices - Finally, there’s the hybrid situation in which you have both
internal and external devices. Most adapters have connectors for both internal and external SCSI
devices - if yours doesn’t have both, you’ll need to see if anybody makes one that will work with
your devices. For adapters that do have both types of connectors, you connect your internal devices
to the ribbon cable and attach the cable to the adapter. Then, you daisy-chain your external devices
off the external port. Finally, you terminate the last device on each chain, leaving the adapter
unterminated.

In addition to running the cable(s) correctly, it is important that you realize that you are limited to
the number of devices you can have on the SCSI channel as well as the maximum length of the SCSI
bus. Generally speaking, the faster the SCSI, the shorter the total length of the bus.

13