ITEC 1000 Introduction to Information Technologies Computer Peripherals – Part I.

ITEC 1000 Introduction to Information Technologies

Computer Peripherals – Part I


Plan

• Storage (hierarchy and terminology)

• Magnetic disks


Storage

• Terminology• Medium

• The technology or product type that holds the data

• Access time• The time to get to the data

• Specified as an average in seconds (e.g., s, ms, µs, ns, etc.)

• Throughput• The rate of transfer for consecutive bytes of data

• Specified in bytes/s (e.g., Kbytes/s, Mbytes/s)


Primarystorage

Storage HierarchyYou should know this hierarchy

Secondarystorage


Terminology• Online storage

• Memory that is accessible to programs without human intervention

• Primary storage and secondary storage are “online”

• Primary storage• Semiconductor technology (e.g., RAM)• Volatile (contents might be lost when powered off )

• Secondary storage• Magnetic technology (e.g., disk drives)• Non-volatile (contents are retained in the absence of power)

Rv: kc


Terminology

• Offline storage• Memory that requires human intervention in order for it

to be accessed by a program (e.g., loading a tape)• Sometimes called “archival storage”

• Direct Access Storage Device (DASD)• Pronounced “dazz-dee”• Term coined by IBM• Distinguishes disks (disk head moves “directly” to the

data) from tapes (tape reel must wind forward or backward to the data: sequential access)

Rv: kc


Plan

• Storage (hierarchy and terminology)

• Magnetic disks


Magnetic Disks

• A magnetic substance is coated on a round surface• The magnetic substance can be polarized in one of

two directions with an electromagnet (“writing data”)

• The electromagnet can also sense the direction of magnetic polarization (“reading data”)

• Similar to a read/write head on a tape recorder (except the information is digital rather than analogue)


Hard Disks

• The platter is “hard” (e.g., aluminum)• Most hard disk drives contain more than one

platter• On most hard disk drives, the disks are “fixed”

(i.e., not removable)• On some hard disk drives, the disks are in a

removable pack (hence, “disk pack”)• Typical speed of rotation: 3600, 5400, 7200 rpm

(rpm = “revolutions per minute”)• Capacities: 500 MB to 10+ TB (terabyte = 240

bytes)


Hard Disk Layout

Platter

Track

Cylinder

Drivemotor

Headmotor

Head, onmoving arm

Block

Sector

Track

Head

Head assembly

1

23

220


Hard Disk Example


Terminology

• Platter• A round surface – the disk – containing a magnetic coating

• Track• A circle on the disk surface on which data are contained

• Head• A transducer attached to an arm for writing/reading data to/from the

disk surface

• Head assembly• A mechanical unit holding the heads and arms• All the head/arm units move together, via the head assembly

• Cylinder• A set of tracks simultaneously accessible from the heads on the head

assembly

YO

U M

US

T K

NO

W T

HIS


Terminology• Drive motor

• The motor that rotates the platters• Typically a DC motor (DC = direct current)• The disk rotates at a fixed speed (e.g., 3600 rpm,

revolutions per minute)

• Head motion• A mechanism is required to move the head assembly in/out• Two possibilities:

• A stepper motor (digital, head moves in steps, no feedback)• A servo motor (analogue, very precision positioning, but requires

feedback)


Terminology

• Sector• That portion of a track falling along a predefined pie-shaped

portion of the disk surface• The number of bytes stored in a sector is the same, regardless

of where the sector is located; thus, the density of bits is greater for sectors near the centre of the disk

• The rotational speed is constant; i.e., constant angular velocity• Thus, the transfer rate is the same for inner sectors and outer

sectors

• Block• The smallest unit of data that can be written or read to/from

the disk (typically 512 bytes)


Locating a Block of Data

Seek Time Latency Time Transfer Rate

Desiredtrack

Seek

Head

TransferLatency

Note: Access time = seek time + latency

123

45 67

8


Terminology

• Seek time• The time for the head to move to the correct track• Specified as an average for all tracks on the disk surface

• Latency time• The time for the correct block to arrive at the head once the head

is positioned at the correct track• Specified as an average, in other words, ½ the period of rotation• Also called “rotational delay”

• Access time is the time “to get to” the data (remember!)• Access time = seek time + latency

• Transfer rate• Same as throughput


Latency Example

• A hard disk rotates at 3600 rpm

• What is the average latency?

Period of rotation = (1 / 3600) minutes= (1 / 3600) 60 seconds= 0.01667 s= 16.67 ms

Average latency = 16.67 / 2 ms= 8.33 ms


Factors Determining Transfer Rate

• Transfer rate can be determined, given…• Rotational speed of the disk platters • Number of sectors per track• Number of bytes per sector


Transfer Rate Example

• Q: Determine the transfer rate, in Mbytes/s, for a hard disk drive, given• Rotational speed = 7200 rpm• Sectors per track = 30• Data per sector = 512 bytes = 0.5 Kbytes

• A: Transfer rate = 7200 x 30 = 216,000 sectors/min

= 216,000 x 0.5 = 108,000 Kbytes/min= 108,000 / 60 = 1,800 Kbytes/s= 1,800 / 210 = 1.76 Mbytes/s


Exercise - Transfer Rate

• Q: Determine the transfer rate, in Mbytes/s, for a hard disk drive, given

• Rotational speed = 7000 rpm• Sectors per track = 32• Data per sector = 1024 bytes

Skip answer Answer


• Q: Determine the transfer rate, in Mbytes/s, for a hard disk drive, given

• Rotational speed = 7000 rpm

• Sectors per track = 32

• Data per sector = 1024 bytes

A: Transfer rate = 7000 x 32 = 224,000 sectors/min

= 224,000 x 1 = 224,000 Kbytes/min= 224,000 / 60 = 3,733 Kbytes/s= 3,733 / 210

= 3.65 Mbytes/s


Track Format

• Format of each track:

dataheadergap gapCRC

SectorPrevious sector Next sector

Inter-blockgap

Inter-blockgap

Note:

CRC stands for “cyclic redundancy check”. It’s the “footer” at the end of each sector. CRC is a sophisticated form of parity for checking that the data read are accurate.

Rv: kc


Figure 9.7 A single data block


Figure 9.8 Header for MS-DOS/Windows disk


Disk Formatting

• The track positions, blocks, headers, and gaps must be established before a disk can be used

• The process for doing this is called “formatting”

• The header, at the beginning of each sector, uniquely identifies the sector, e.g., by track number and sector number


Disk Controller• Interface between the disk drive and the

system is known as a “disk controller”

• A primary function is to ensure data read/write operations are from/to the correct sector

• Since data rate to/from the disk is different than data rate to/from system memory, “buffering” is needed

Rv: kc


Buffering

System Diskcontroller

DiskRAM Buffer

(RAM)

1. Read data from disk into a buffer in the disk controller

2. Transfer data from buffer to system RAM (Note: this is a DMA operation)

Example: Reading data from a disk


Multi-block Transfers (1 of 2)

• The smallest transfer is one block (e.g., 512 bytes)• However, often multi-block transfers are required• The inter-block gap provides “time” for the controller

electronics to adjust from the end of one sector to the beginning of the next

• “time” may be needed for a few reasons:• Compute and/or verify the CRC bytes• Switch circuits from read mode to write mode

• During a write operation the header is “read” but the data are “written”• (Remember, the header is only “written” during formatting.)

• Perform a DMA operation


Multi-block Transfers (2 of 2)

• Sometimes, sectors simply cannot be read or written consecutively

• There is not enough time (see preceding slide)

• The result is lost performance because the disk must undergo a full revolution to read the next sector

• The solution: interleaving


Interleaving

• Rather than numbering blocks consecutively, the system skips one or more blocks in its numbering

• This allows multi-block transfers to occur as fast as possible

• Interleaving minimizes lost time due to latency• Interleaving “factor” (see next slide) is established

when the disk is formatted• Can have a major impact on system performance

A “must know” item


Interleaving Examples

21 3 54 6 87 9

1 2 3 4 5

1 2 3

1:1 etc.

etc.

etc.

2:1

3:1

Factor


2:1 Interleaving

1

2 6

7

3

84

9

5


File System Considerations• There is no direct relationship between the size and

physical layout of blocks on a disk drive and the size and organization of files on a system

• File system• Determines the organization of information on a computer• Performs logical-to-physical mapping of information• A file system is part of each and every operating system

• Logical mapping• The way information is perceived to be stored

• Physical mapping• The way information is actually stored


Alternative Technologies (1 of 3)

• Removable hard disks• Also called “disk packs”• A stack of hard disks enclosed in a metal or plastic removable cartridge• Advantages

• High capacity and fast, like hard disk drives• Portable, like floppy disks

• Disadvantage• Expensive



• Fixed heads• Fewer tracks but eliminates seek time

Disk Spindle Moving head

Fixed heads



• R.A.I.D. = Redundant array of inexpensive disks• A category of disk drive that employs two or

more drives in combination for fault tolerance and performance

• Frequently used on servers, but not generally used on PCs

• There are a number of different R.A.I.D. “levels” (next slide)


R.A.I.D. Levels (1 of 2)

• Level 0• Provides “data striping” (spreading out blocks of each file

across multiple disks)• No redundancy• Improves performance, but does not deliver “fault tolerance”

• Level 1• Provides “data mirroring”: (a.k.a.: “shadowing”)• Data are written to two duplicate disks simultaneously• If one drive fails, the system can switch to the other without

loss of data or service• Delivers fault tolerance

Rv kc


R.A.I.D. Levels (2 of 2)

• Level 3• Same as level 0, but also reserves one dedicated

disk for error correction data• Good performance, and some level of fault

tolerance

• Level 5• Data striping at the byte level and stripe error

correction information• Excellent performance, good fault tolerance


Terminology

• Fault tolerance• The ability of a computer system to respond gracefully

to unexpected hardware or software failure

• Many levels of fault tolerance• E.g., the ability to continue operating in the event of a power

failure

• Some systems “mirror” all operations• Every operation is performed on two or more duplicate

systems, so if one fails, another can take over


Terminology

• Data mirroring (also: shadowing)• A technique in which data are written to two

duplicate disks simultaneously• If one disk fails, the system can instantly switch

to the other disk without loss of data or service• Used commonly in on-line database systems

where it is critical that data are accessible at all times


Terminology

• Data striping• A technique for spreading data over multiple

disks• Speeds operations that retrieve data from disk

storage• Data are broken into units (blocks) and these

are spread across the available disks• Implementations allow selection of data units

size, or stripe width


CD-ROM

• CD-ROM stands for “compact disc, read-only memory”

• Evolved from audio CDs

• Disk size: 120 mm (5¼”)

• Capacity: 750 MB


Operation

• Uses light generated by lasers to record and retrieve information

• Information is stored by varying the light reflectance characteristics of the medium

• Available in read-only (CD-ROM) and read/write formats


Layout of a CD-ROM versus a standard disk

75 sectors/second60 sec/min


CD-ROM vs. Magnetic Disk

CD-ROM Magnetic Disk

• One spiral track (3 miles long!)

• Multiple tracks of concentric circles

• Constant bit density • Variable bit density

• Disk speed varies (CLV, constant linear velocity)

• Disk speed constant (CAV, constant angular velocity)

• Constant transfer rate • Constant transfer rate

• Capacity: 750 MB • Capacity: varies


CD-ROM Data Organization

• 270,000 blocks of 2048 bytes each (typically)

• 270,000 2048 = 552,960,000 bytes• Extensive error checking and correction

(e.g., bad regions of the disk flagged)• Substantial overhead for error correction

and identifying blocks• Capacity can be as high as 750 MB


Pits and Lands (1 of 2)

• Data are stored as “pits” and “lands”• These are burned into a master disk by a high

powered laser• Master disk is reproduced mechanically by a

stamping process. ( Like a coin, sort of. )• Data surface is protected by a clear coating• Data are read by sensing the reflection of laser light

• A pit scatters the light• A land reflects the light


Pits and Lands (2 of 2)

Laser

Land

Reflectedlight

LaserLaser

Pit

Scatteredlight


CD-ROM Read Process

Laserdiode

Prism

Light detector

Land Pit

Transparentprotective layer

More detail


Magneto Optical

• Disk may be written, read, and rewritten• Write process is preformed at high temperature• Combines features of optical and magnetic

technology• Data are stored as a magnetic charge on the disk

surface• During reading, the polarity of the reflected light

is sensed (not the intensity)


http://en.wikipedia.org/wiki/DVD-R

ITEC 1000 Introduction to Information Technologies Computer Peripherals – Part I.

Documents

disks disk head

hard disk drives

kc slide

disk surface head assembly

mbytess slide

analogue slide

readwrite head

magnetic substance