Part 2Computer Systems

(Aux and I/O)Hardware

(Text No. 1 Chapter 2)

HARDWARE

• 2.4 Auxiliary storage devices

• 2.5 Input/output architecture and devices

Auxiliary Storage Devices

• Types and characteristics of storage devices– Main storage unit

– Auxiliary storage devices

Magnetic Tape

Magnetic Tape

• Type 0– Original ferric-oxide tape– Very rarely seen these days

• Type 1– Standard ferric-oxide tape– Also referred to as "normal bias"

Magnetic Tape

• Type 2– "Chrome" or CrO2 tape– The ferric-oxide particles are mixed with

chromium dioxide

• Type 4– "Metal" tape– Metallic particles rather than metal-oxide

particles are used in the tape

Magnetic Disk Unit

• Devices that store data using magnetic disks.

• Most widely used auxiliary storage device

• Magnetic disks for personal computers or workstations are also called fixed disks or hard disks but the mechanism is the same.

Magnetic Disk Unit

Opening Up Hard Disk

Platters and Heads

Inside Hard Disk

Magnetic Media• Tracks

– Data recorded along rings called ‘tracks’

– Length of the outer tracks are larger than that of the inner tracks differ but:

• Storage capacity is the same• Storage density increases from

outer to inner

Magnetic Media

Magnetic Media

• Cylinders– In magnetic disk units

• Multiple magnetic disks

• Groups of tracks with the same radius on each of the disks is set as one data storage area called a ‘cylinder’

• Makes for more efficient data retrieval

• Multiple fixed arm magnetic heads

Magnetic Media

Inside Hard disk

Magnetic Disk Structure

Magnetic Media

Storage Capacity

= storage capacity of 1 track *

number of tracks per cylinder *

number of cylinders of the magnetic disk

Calculating Storage Capacity• Specifications

Number of cylinders: 800 Number of tracks per cylinder: 19 Storage capacity per track: 20,000 bytes

Calculating Storage Capacity• Specifications

Number of cylinders: 800 Number of tracks per cylinder: 19 Storage capacity per track: 20,000 bytes

Calculations• Storage capacity per cylinder

20,000 bytes * 19 tracks = 380,000 bytes/cylinder = app. 380KB

• Storage capacity of disk 380KB * 800 cylinders = 304,000 KB = app. 304 MB

Magnetic Disk Unit

• Recording Type– Variable type

• Data reading, writing performed on block basis.• Read/write can be started from any track position• Used in magnetic disks• IBG = Inter-Block Gap

Calculating Storage Capacity (Blocking)

• Calculate the number of cylinders required when 80,000 records of 200 bytes each are stored in a sequential access file of 10 records/block.

• Blocking cannot be extended over multiple tracks.

• Number of cylinders : 400• Number of tracks/cylinder : 19• Storage capacity/track : 20,000 bytes• Inter-block gap (IBG) : 120 bytes

Calculating Storage Capacity (Blocking)

• Number of blocks for the whole file = number of records/blocking factor– 80,000 / 10 = 8,000 blocks

• Length of 1 block, including the IBG– 200 bytes/record * 10 records/block + 120 bytes/block = 2,120 bytes/block

• Number of blocks on 1 track– 20,000 bytes/track / 2,120 bytes/block = app. 9.43 blocks/track– Because blocking cannot be extended over multiple tracks, floor(9.43) = 9 blocks/track

• Number of tracks for the whole file– 8,000 blocks / 9 blocks/track = app. 888.88 tracks– Ceil(888.88) = 889 tracks

• Number of cylinders required to write the whole file = number of tracks / number of tracks/cylinder– 889 / 19 = app. 46.78 cylinders– Ceil (46.78) = 47 cylinders.

Magnetic Media

• Structure and operation principles– Sector type

• Each track divided into approximately 20 small sectors.

• Reading/writing specified with sector number of selected track.

• Used in hard disks and floppy disks.

Magnetic Disk Structure

Magnetic Disk

• Parity check– Magnetic head reads/writes data to track bit by bit

appending an extra bit for parity check.

• Defragmentation– Data written on hard disks are not contiguous

resulting in slow access time.

– Defragmentation will arrange data contiguously thereby speeding up read/write time

Magnetic Disk

• Access is the generic term for the act of reading specific data from the magnetic disk and writing it on a specific cylinder or track. Access time is calculated through the addition of the following:– Seek time– Search time– Data transfer time

Magnetic Disk

• Seek time– Refers to the time a program or

device takes to locate a particular piece of data. For disk drives, the terms seek time and access time are often used interchangeably. Technically speaking, however, the access time is often longer than seek time because it includes a brief latency period.

Magnetic Disk

• Search time or latency– Lapse of time until

target data reaches the magnetic head position

Magnetic Disk

• Data transfer time– Time elapsed between when the magnetic head

data access starts and when the transfer is completed

Access Time

• Access Time of Magnetic Disk Unit = Average Seek Time

+ Average Search Time

+ Data Transfer Time

Calculating Access Time

• Specifications• Capacity/track : 45,000 bytes

• Rotation speed : 2,500 rpm

• Average seek time : 10 s

• Find the access time (ms) for 15,000 bytes of data

• Calculation• Average search time

– Revolution speed = 3,000 rpm

– 3,000 rev in 60 sec

– n(revolutions) in 1 second = 3000 / 60 = 50 rev/sec

– 1 rev in (1/50)sec = 0.02 sec/revolution = 20ms/rev

– Average search time = 20ms / 2 = 10ms

Calculating Access Time

• Data Transfer Speed– In 1 revolution, the information contained in 1 track passes

through the magnetic disk head– The disk makes 50 rev/sec– Data transfer speed = 50 tracks/sec * 15,000 bytes/track =

750,000 = 750 * 103 bytes/sec– Data transfer time for 9,000 bytes of data =

(9*103)/(750*103) = 0.012 sec = 12 ms• Access Time = Average Seek Time + Average Search

Time + Data Transfer Time– 20ms + 10ms + 12ms = 42 ms

Floppy Disk

Track – Concentric ring of data on a side of a disk.

Sector – A subset of a track, similar to wedge or a slice of pie.

Floppy Disk• Recording method is sector type.

– Within the outer protective casing is a circular flexible disk, hence “floppy”

– Low priced storage unit and easily transported, it is widely used.

Floppy Disk Structure

Floppy Disk

Read/Write• Located on both sides of a diskette• The heads are not directly opposite each other• The same head is used for reading/writing, while a

second wider head is used for erasing a track just prior to it being written.

Floppy Disk

Stepper Motor• Makes a precise number of stepped revolutions to

move the read/write head assembly to the proper track position

• The read/write head assembly is fastened to the stepper motor shaft

Floppy Disk• Storage capacity =

Storage capacity per sector *

Number of sectors per track *

Number of tracks per side *

Number of sides (One side or both sides)

Calculating Floppy Disk Capacity

• Specifications– Sides available for use: 2 sides– Track number/side: 80 tracks– Sector number/track: 9 sectors– Storage capacity/sector: 1,024 bytes

• Storage capacity of 1 track– 1,024 bytes/sector 9 sectors/track = 9,216 bytes/track

• Storage capacity of 1 side is as follows:– 9,216 bytes/track 80 tracks = 737,280 bytes = app. 737kB

• Both sides used:• 737kB 2 = 1,474kB = app. 1.44MB

High Capacity Floppy Disks

– 100 MB or greater capacity

– Store large files such as graphics, audio, or video

– Used for backups

– SuperDisk drive

– Zip drive

Optical Disk (CD, DVD) Unit

• Optical disk units– Store/save image processing data of extremely

large volume or as storage devices of large volume packaged software.

– These devices can store large volumes of data through a mechanism that reads out information using light reflection

– Widely used form of multimedia storage and distribution

CD-ROM

• Compact disc read-only memory• Can contain text, graphics, and video as

well as sound• Cannot be erased or modified• Use CD-ROM drive or CD-ROM player to

read• Holds about 650-850 MB• Used to distribute software

Optical Disk (CD, DVD) Unit

• Music (Audio) CD (CDA)• CD-ROM

– CD-G (CD-Graphic) for image data– CD-I (CD-Interactive) for interactive applications– Photo-CD

• CD-R (CD-Recordable)• CD-RW (CD-Rewritable)• Multi-session

CD-ROM Structure

CD Surface

Reading from CD-ROM

Disney World on CD-ROM

CD-ROM Performance

• Seek time– Slow compared to magnetic disks due to heavy lens in

the read head and the spiral structure

• Transfer rate– Expressed in numeric values that represent how much

data can be transferred in comparison with audio CDs. – Audio CD player : 150kB/s– CD-ROM units with transfer speeds 2 times or 3 times

as fast as the transfer rate for audio CDs began to be developed and today the transfer rate has reached levels of 50x.

Optical Disk Specifications• Red Book

– Basic CD standard. Describes physical specifications of the CD

• Yellow Book– Basic CDROM standard. Physical specifications of the

CDROM

• Green Book– Defines physical structure of CD-I (CD Interactive)

• Orange Book– Defines physical structure of the CD-R (CD-

Recordable)

DVD

• DVD (Digital Versatile Disk or Digital Video Disk)– Capable of storing up to 2 hours (or more,

depending on standards used) of animated images and audio data

– Can be thought of as layering CD-ROMs one on top of the other

Hollywood on DVD

DVD Structure• Uses MPEG2 compression technology

– High quality audio/video

• Variations– DVD-ROM

– DVD-R

– DVD-RAM

• Regional encoding

DVD Surface

DVD Recording

DVD Capacity

• Single layer single sided recording: 4.7 GB

• Dual layer single sided recording: 8.5 GB

• Single layer dual sided recording: 9.4 GB

• Dual layer dual sided recording: 17 GB

CD versus DVD

Specification CD DVD

Track Pitch 1600 nanometers 740 nanometers

Min Pit Length (Single layer) 830 nanometers 400 nanometers

Min Pit Length (Double layer) NA 440 nanometers

Note that double layer is only applicable to DVD.

Semiconductor Disk Unit

• Storage unit of high speed and large capacity• Uses flash memories and other devices• Usually used in high-end mainframe computers as

a storage unit positioned between the main storage unit and the auxiliary storage devices.

• Several G bytes of storage capacity• Access time is 1/100th that of the magnetic disk

unit

Flash Memory

RAID

• High reliability and high performance of storage becoming increasingly important

• Especially in real-time networked environments• Redundant Arrays of Inexpensive Disks

– Fault-tolerant– Use from 2 to 32 disks in an array– Most commonly used RAID 0, RAID 1 and RAID 5– Rapid data recovery (rollback) in the event of a disk

failure• Can be implemented in hardware which is the

preferred method or software which is cheaper

RAID 0 • Striping

– Data is stripped among all available disks

– Between 2 and 32 disks are used. Fastest read/write

– No redundancy, hence no fault-tolerance• One disk failure, all data in array are lost

– Useful for video production and editing, image editing, and any application requiring high bandwidth

RAID 1• Disk Mirroring/Duplexing

– Uses 2 drives with same capacity where same content recorded on two hard disks concurrently

– Useful for critical mission applications

– Slowest of the RAID levels

– Fault tolerant - if one disk fails, data can be recovered from the other disk

– Single point of failure. If controller fails, both disks will be lost.

RAID 2

– Bit interleaved data is distributed over multiple hard disks with parity and error correction information

– No commercial implementations exist / not commercially viable

Hamming Code ECC

RAID 3

– Bit interleaved data is distributed over multiple hard disks, and parity and error correction information stored on a dedicated hard disk

– Controller design is fairly complex– Useful for video production and live streaming

RAID 4

– Block (by sector) data is distributed over multiple hard disks, and parity and error correction information stored on a dedicated hard disk

– Difficult and inefficient data rebuild in the event of disk failure

RAID 5• RAID 5 (Striping with parity)

– Most used for fault tolerance and speed

– Both block (by sector) data and parity and error correction information are distributed over multiple hard disks

– Most complex controller design

– Most commonly adopted in today's servers

– Uses between 3 to 32 disks

RAID 6

– Also known as fault tolerant system

– Hard disks form a matrix for row and column parity – faulty disk can be identified and replaced

– Very poor write performance

RAID 7

– Heterogeneous system supports multiple hosts

– One vendor (Storage Computer Corporation) proprietary solution

– Extremely high cost per MB

I/O Architecture and Devices

• Input/output control method– Bus– Control bus– Data bus

• Direct control method

• DMA (Direct Access Memory) method

I/O Architecture and Devices

I/O Control Method

• Bus– A collection of wires through which data is transmitted from one

part of a computer to another – All buses consist of two parts -- an address bus and a data bus.

The data bus transfers actual data whereas the address bus transfers information about where the data should go

– The size of a bus, known as its width, determines how much data can be transmitted at one time. For example, a 16-bit bus can transmit 16 bits of data, whereas a 32-bit bus can transmit 32 bits of data.

– Every bus has a clock speed measured in MHz

Bus

Bus

Input / Output Architecture and Devices

Input / Output Control Method• Address Bus

– A collection of wires connecting the CPU with main memory that is used to identify particular locations (addresses) in main memory.

– The width of the address bus (that is, the number of wires) determines how many unique memory locations can be addressed

Input / Output Architecture and Devices

Input / Output Control Method

• Control Bus– Connects control unit and RAM.

– Used for transmission of instruction signals to RAM unit from control unit

Input / Output Architecture and Devices

Input / Output Control Method• Data Bus

– Connects the main storage unit and the processor

– Used to exchange data.

– Only this bus is used to exchange data between the main storage unit and the processor in both directions.

Input / Output Architecture and Devices

Input / Output Control Method• Direct Control Method

– CPU directly controls input/output operations • Data exchange is performed through the processor

• Processor cannot perform the next action till the input/output operation is completed

• Inefficient; this method is not widely used

Input / Output Architecture and Devices

Input / Output Control Method• DMA (Direct Memory Access)

– a technique for transferring data from main memory to a device without passing it through the CPU.

– Fast data transfer as the CPU is not involved

– Usually used for real applications, widely used in PC environments

Input / Output Architecture and Devices

• Input/output interfaces– Serial interface

• RS-232C• USB• IEEE 1394

– Parallel interface• Centronics interface• SCSI• GPIB

Summary of I/O Interfaces