Chapter Seven : Device Management
• System Devices Paper Storage Media
• Sequential Access Storage
Magnetic Tape Storage
• Direct Access Storage
• Components of I/O Magnetic Disk Storage
• Communication Among
Devices Optical Disc Storage
• Management of I/O
Device Management Functions
• Track status of each device (such as tape drives, disk
drives, printers, plotters, and terminals).
• Use preset policies to determine which process will get a
device and for how long.
• Allocate the devices.
• Deallocate the devices at 2 levels:
– At process level when I/O command has been executed & device is
– At job level when job is finished & device is permanently
• Differences among system’s peripheral devices are a
function of characteristics of devices, and how well they’re
managed by the Device Manager.
• Most important differences among devices
– Degree of sharability.
• By minimizing variances among devices, a system’s
overall efficiency can be dramatically improved.
• Assigned to only one job at a time and serve that job for
entire time it’s active.
– E.g., tape drives, printers, and plotters, demand this
kind of allocation scheme, because it would be
awkward to share.
• Disadvantage -- must be allocated to a single user for
duration of a job’s execution.
– Can be quite inefficient, especially when device isn’t
used 100 % of time.
• Assigned to several processes.
– E.g., disk pack (or other direct access storage device)
can be shared by several processes at same time by
interleaving their requests.
• Interleaving must be carefully controlled by Device
• All conflicts must be resolved based on predetermined
policies to decide which request will be handled first.
• Combination of dedicated devices that have been
transformed into shared devices.
– E.g, printers are converted into sharable devices
through a spooling program that reroutes all print
requests to a disk.
– Output sent to printer for printing only when all of a
job’s output is complete and printer is ready to print out
– Because disks are sharable devices, this technique can
convert one printer into several ―virtual‖ printers, thus
improving both its performance and use.
Sequential Access Storage Media
• Magnetic tape used for secondary storage on early
computer systems; now used for routine archiving &
storing back-up data.
• Records on magnetic tapes are stored serially, one after
• Each record can be of any length.
– Length is usually determined by the application program.
• Each record can be identified by its position on the tape.
• To access a single record, tape is mounted & ―fast-
forwarded‖ from its beginning until locate desired position.
• Data is recorded on 8 parallel
tracks that run length of tape.
• Ninth track holds parity bit used Parity
for routine error checking.
• Number of characters that can be
recorded per inch is determined • •
by density of tape (e.g., 1600 or •
6250 bpi). Characters •
Storing Records on Magnetic Tapes
• Can store records individually or grouped into blocks.
– If individually, each record is separated by a space to indicate its
starting and ending places.
– If blocks, then entire block is preceded by a space and followed by
a space, but individual records are stored sequentially within block.
• Interrecord gap (IRG) is gap between records about 1/2
inch long regardless of the sizes of the records it separates.
• Interblock gap (IBG) the gap between blocks of records;
still 1/2 inch long.
Pros & Cons of Blocking
• Fewer I/O operations are needed because a single READ
command can move an entire block (physical record that
includes several logical records) into main memory.
• Less tape is wasted because size of physical record
exceeds size of gap.
• Overhead and software routines are needed for blocking,
deblocking, and record keeping.
• Buffer space may be wasted if you need only one logical
record but must read an entire block to get it.
Transfer Rates & Speeds
• Block size set to take advantage of transfer rate.
• Transfer rate -- density of the tape, multiplied by the tape
transport speed (speed of the tape)
transfer rate = density * transport speed
• If transport speed is 200 inches per second, at 1600 bpi, a
total of 320,000 bytes can be transferred in one second,
– Theoretically optimal size of a block is 320,000 bytes.
– Buffer must be equivalent.
Magnetic Tape Access Times Vary Widely
Benchmarks Access time
Maximum access 2.5 minutes
Average access 1.25 minutes
Sequential access 3 milliseconds
• Variability makes magnetic tape a poor medium for routine
secondary storage except for files with very high
Direct Access Storage Devices (Random
Access Storage Devices)
• Direct access storage devices (DASDs)-- any devices that
can directly read or write to a specific place on a disk.
• Two major categories:
• DASD with fixed read/write heads
• DASD with movable read/write heads.
• Although variance in DASD access times isn’t as wide as
with magnetic tape, location of specific record still has a
direct effect on amount of time required to access it.
• Magnetically recordable drums.
• Resembles a giant coffee can
covered with magnetic film and
formatted so the tracks run
• Data is recorded serially on
each track by the read/write
head positioned over it.
• Fixed-head drums were very
fast but also very expensive,
and they did not hold as much
data as other DASDs.
Fixed Head Disks
• Fixed-head disks -- each
disk looks like a phonograph Rotation
• Covered with magnetic film
that has been formatted,
usually on both sides, into
• Each circle is a track. Data
is recorded serially on each
track by the fixed read/write
head positioned over it.
• One head for each track.
Pros & Cons of Fixed Head Disks
• Very fast—faster than movable-head disks.
• High cost.
• Reduced storage space compared to a moveable-head disk
– because tracks must be positioned farther apart to accommodate
width of the read/write heads.
Movable-Head Drums and Disks
• Movable-head drums have only a few read/write heads that
move from track to track to cover entire surface of drum.
– Least expensive device has only 1 read/write head for entire drum
– More conventional design has several read/write heads that move
• One read/write head that floats over the surface of the disk.
• Disks can be individual units (used with many PCs) or part
of a disk pack (a stack of disks).
• It’s slower to fill a disk pack surface-by-surface than to fill
it up track-by-track.
• If fill Track 0 of all surfaces, got virtual cylinder of data.
– Are as many cylinders as there are tracks.
– Cylinders are as tall as the disk pack.
• To access any given record, system needs:
– Cylinder number, so arm can move read/write heads to it.
– Surface number, so proper read/write head is activated.
– Record number, so read/write head know when to begin reading or
Optical Disc Storage (CD-ROM)
• Optical disc drives uses a laser beam to read and write to
• Optical disc drives work in a manner similar to a magnetic
– Head on an arm that moves forward and backward across the disc.
• Uses a high-intensity laser beam to burn pits (indentations)
and lands (flat areas) in disc to represent ones and zeros,
Concentric Tracks vs. Spiraling Tracks
• Magnetic disk consists of concentric tracks of sectors and
it spins at a constant speed (constant angular velocity).
– Because sectors at outside of disk spin faster past
read/write head than inner sectors, outside sectors are
much larger than sectors located near center of disk.
• An optical disc consists of a single spiraling track of same-
sized sectors running from center to rim of disc.
– Allows many more sectors & much more data to fit on
optical disc compared to magnetic disk of same size.
Measures of Performance
for Optical Disc Drives
• Sustained data-transfer rate -- speed at which massive
amounts of data can be read from disc.
– Measured in bytes per second (such as Mbps).
– Crucial for applications requiring sequential access.
• Average access time -- average time required to move
head to a specific place on disc.
– Expressed in milliseconds (ms).
• Cache size -- hardware cache acts as a buffer by
transferring blocks of data from the disc
– Anticipates user may want to reread some recently retrieved info.
– Act as read-ahead buffer, looking for next block of info on disc.
• CD-ROM -- first commonly used optical storage DASD.
• Stores very large databases, reference works, complex
games, large software packages, system documentation,
and user training material.
• CD-ROM jukeboxes (autochangers or libraries) are
capable of handling multiple discs and networked to
distribute multimedia and reference works to distant user.
CD-Recordable Technology (CD-R)
• CD-R drives record data on optical discs using a write-
• WORM (write once, read many).
• Only a finite amount of data can be recorded on each disc
and, once data is written, it can’t be erased or modified.
• It has an extremely long shelf life.
CD-Rewritable Technology (CD-RW)
• CD-RW drives can read a standard CD-ROM, CD-R and
• CD-RW discs can be written and rewritten many times by
focusing a low-energy laser beam on surface, heating
media just enough to erase pits that store data and restoring
recordable media to its original state.
• Useful for storing large quantities of data and for sound,
graphics, and multimedia applications.
Digital Video Disc (DVD) Techonolgy
• DVD uses infrared laser to read disc (holds equivalent of
13 CD-ROM discs).
• By using compression technologies, has more than enough
space to hold a 2-hour of movie with enhanced audio.
– Single layered DVDs can hold 4.7 GB
– Double-layered disc can hold 8.5 GB on each side of the disc.
DVDs are used to store music, movies, and multimedia
• DVD-RAM is a writable technology that uses a red laser
to read, modify, and write data to DVD discs.
Three Factors Contribute To Time
Required To Access a File
• Seek time -- time required to position the read/write head
on the proper track. (Doesn’t apply to devices with fixed
– Slowest of the three factors
• Search time (rotational delay) -- time it takes to rotate
DASD until requested record is under read/write head.
• Transfer time -- when data is actually transferred from
secondary storage to main memory.
Access Time For Fixed-Head Devices
• Fixed-head devices can access a record by knowing its
track number and record number.
• Total amount of time required to access data depends on:
– Rotational speed is constant within each device (although it varies
from device to device)
– Position of record relative to position of the read/write head.
search time (rotational delay)
+ transfer time (data transfer)
Example of Access Time For
• How long will it take to access a record?
• Typically, one complete revolution takes 16.8 ms, so
average rotational delay is 8.4 ms.
• Data transfer time varies from device to device, but a
typical value is 0.00094 ms per byte
– size of record dictates this value.
• For example, it takes 0.094 ms (almost 0.1 ms) to transfer
a record with 100 bytes.
Access Time For Movable-Head Devices
• Movable-head DASDs adds time required to move arm
into position over the proper track (seek time).
seek time (arm movement)
search time (rotational delay)
+ transfer time (data transfer)
• Seek time is the longest and several strategies have been
developed to minimize it.
Components of the I/O Subsystem
Control Unit 1
Channel 1 Disk 2
Control Unit 2 Disk 3
Control Unit 3 Tape 3
Channel 2 Tape 4
Control Unit 4
I/O Subsystem : I/O Channel
• I/O Channel -- keeps up with I/O requests from CPU and
pass them down the line to appropriate control unit.
– Programmable units placed between CPU and control unit.
– Synchronize fast speed of CPU with slow speed of the I/O device.
– Make it possible to overlap I/O operations with processor
operations so the CPU and I/O can process concurrently.
• Use channel programs that specifies action to be
performed by devices & controls transmission of data
between main memory & control units.
• Entire path must be available when an I/O command is
I/O Subsystem : I/O Control Unit
• I/O control unit interprets signal sent by channel.
– One signal for each function.
• At start of I/O command, info passed from CPU to
– I/O command (READ, WRITE, REWIND, etc.)
– Channel number
– Address of physical record to be transferred (from or to secondary
– Starting address of a memory buffer from which or into which
record is to be transferred
Device Manager Must
• Know which components are busy
and which are free.
Solved by structuring
units • Be able to accommodate requests
that come in during heavy I/O traffic.
• Accommodate disparity of speeds
between CPU and I/O devices.
Handled by ―buffering‖
records & queueing
Communication Among Devices
• Each unit in I/O subsystem can finish its operation
independently from others.
• CPU is free to process data while I/O is being performed,
which allows for concurrent processing and I/O.
• Success of operation depends on system’s ability to know
when device has completed operation.
– Uses a hardware flag that must be tested by CPU.
Hardware Flag Used To Communicate When A
Device Has Completed An Operation
• Composed made up of three bits.
– Each bit represents a component of I/O subsystem.
– One each for channel, control unit, and device.
• Resides in the Channel Status Word (CSW)
– In a predefined location in main memory and contains
info indicating status of channel.
• Each bit is changed from zero to one to indicate that unit
has changed from free to busy.
Testing the Flag : Polling or Interrupts
• Polling uses a special machine instruction to test flag.
– CPU periodically tests the channel status bit (in CSW).
• Major disadvantage with this scheme is determining how
often the flag should be polled.
– If polling is done too frequently, CPU wastes time
testing flag just to find out that channel is still busy.
– If polling is done too seldom, channel could sit idle for
long periods of time.
• Use of interrupts is a more efficient way to test flag.
• Hardware mechanism does test as part of every machine
instruction executed by CPU.
• If channel is busy flag is set so that execution of current
sequence of instructions is automatically interrupted.
• Control is transferred to interrupt handler, which resides
in a predefined location in memory.
• Some sophisticated systems are equipped with hardware
that can distinguish between several types of interrupts.
Direct Memory Access (DMA)
• I/O technique that allows a control unit to access main
• Once reading or writing begins, remainder of data can be
transferred to and from memory without CPU intervention.
• To activate this process CPU sends enough info to control
unit to initiate transfer of data
• Then CPU goes to another task while control unit
completes transfer independently.
• This mode of data transfer is used for high-speed devices
such as disks.
• Buffers are temporary storage areas residing in convenient
locations throughout system: main memory, channels, and
• Used extensively to better synchronize movement of data
between relatively slow I/O devices & very fast CPU.
• Double buffering --2 buffers are present in main memory,
channels, and control units.
– While one record is being processed by CPU another can be read
or written by channel
Management of I/O Requests
• Device Manager divides task into 3 parts, with each
handled by specific software component of I/O subsystem.
• I/O traffic controller watches status of all devices, control
units, and channels.
• I/O scheduler implements policies that determine
allocation of, and access to, devices, control units, and
• I/O device handler performs actual transfer of data and
processes the device interrupts.
I/O Traffic Controller
• Monitors status of every device, control unit, and channel.
– Becomes more complex as number of units in I/O subsystem
increases and as number of paths between these units increases.
• Three main tasks: (1) it must determine if there’s at least 1
path available; (2) if there’s more than 1 path available, it
must determine which to select; and (3) if paths are all
busy, it must determine when one will become available.
• Maintains a database containing status and connections for
each unit in I/O subsystem, grouped into Channel Control
Blocks, Control Unit Control Blocks, and Device Control
Traffic Controller Maintains Database For
Each Unit In I/O Subsystem
• I/O scheduler performs same job as Process Scheduler-- it
allocates the devices, control units, and channels.
• Under heavy loads, when # requests > # available paths,
I/O scheduler must decide which request satisfied first.
• I/O requests are not preempted: once channel program has
started, it’s allowed to continue to completion even though
I/O requests with higher priorities may have entered queue.
– Feasible because programs are relatively short (50 to 100 ms).
I/O Scheduler - 2
• Some systems allow I/O scheduler to give preferential
treatment to I/O requests from ―high-priority‖ programs.
– If a process has high priority then its I/O requests also
has high priority and is satisfied before other I/O
requests with lower priorities.
• I/O scheduler must synchronize its work with traffic
controller to make sure that a path is available to satisfy
selected I/O requests.
I/O Device Handler
• I/O device handler processes the I/O interrupts, handles
error conditions, and provides detailed scheduling
algorithms, which are extremely device dependent.
• Each type of I/O device has own device handler algorithm.
– first come first served (FCFS)
– shortest seek time first (SSTF)
– SCAN (including LOOK, N-Step SCAN, C-SCAN, and C-LOOK)
• Every scheduling algorithm should :
– Minimize arm movement
– Minimize mean response time
– Minimize variance in response time
First Come First Served (FCFS) Device
• Simplest device-scheduling algorithm:
• Easy to program and essentially fair to users.
• On average, it doesn’t meet any of the three goals of a seek
• Remember, seek time is most time-consuming of functions
performed here, so any algorithm that can minimize it is
preferable to FCFS.
Shortest Seek Time First (SSTF) Device
• Uses same underlying philosophy as shortest job next
where shortest jobs are processed first & longer jobs wait.
• Request with track closest to one being served (that is, one
with shortest distance to travel) is next to be satisfied.
• Minimizes overall seek time.
• Favors easy-to-reach requests and postpones traveling to
those that are out of way.
SCAN Device Scheduling Algorithm
• SCAN uses a directional bit to indicate whether the arm is
moving toward the center of the disk or away from it.
• Algorithm moves arm methodically from outer to inner
track servicing every request in its path.
• When it reaches innermost track it reverses direction and
moves toward outer tracks, again servicing every request
in its path.
LOOK (Elevator Algorithm) : A Variation
• Arm doesn’t necessarily go all the way to either edge
unless there are requests there.
• ―Looks‖ ahead for a request before going to service it.
• Eliminates possibility of indefinite postponement of
requests in out-of-the-way places—at either edge of disk.
• As requests arrive each is incorporated in its proper place
in queue and serviced when the arm reaches that track.
Other Variations of SCAN
• N-Step SCAN -- holds all requests until arm starts on way
back. New requests are grouped together for next sweep.
• C-SCAN (Circular SCAN) -- arm picks up requests on its
path during inward sweep.
– When innermost track has been reached returns to outermost track
and starts servicing requests that arrived during last inward sweep.
– Provides a more uniform wait time.
• C-LOOK (optimization of C-SCAN) --sweep inward stops
at last high-numbered track request, so arm doesn’t move
all the way to last track unless it’s required to do so.
– Arm doesn’t necessarily return to the lowest-numbered track; it
returns only to the lowest-numbered track that’s requested.
Which Device Scheduling Algorithm?
• FCFS works well with light loads, but as soon as load
grows, service time becomes unacceptably long.
• SSTF is quite popular and intuitively appealing. It works
well with moderate loads but has problem of localization
under heavy loads.
• SCAN works well with light to moderate loads and
eliminates problem of indefinite postponement. SCAN is
similar to SSTF in throughput and mean service times.
• C-SCAN works well with moderate to heavy loads and has
a very small variance in service times.
Search Strategies: Rotational Ordering
• Rotational ordering -- optimizes search times by ordering
requests once read/write heads have been positioned.
– Nothing can be done to improve time spent moving
read/write head because it’s dependent on hardware.
• Amount of time wasted due to rotational delay can be
– If requests are ordered within each track so that first
sector requested on second track is next number higher
than one just served, rotational delay is minimized.
Redundant Array of Inexpensive Disks
• RAID is a set of physical disk drives that is viewed as a
single logical unit by OS.
• RAID assumes several smaller-capacity disk drives
preferable to few large-capacity disk drives because, by
distributing data among several smaller disks, system can
simultaneously access requested data from multiple drives.
• System shows improved I/O performance and improved
data recovery in event of disk failure.
• RAID introduces much-needed concept of redundancy to
help systems recover from hardware failure.
• Also requires more disk drives which increase hardware
Six standard levels of RAID fall into 4 categories. Each
offers a unique combination of advantages.
• access time • I/O device handler
• blocking • I/O scheduler
• buffers • I/O subsystem
• Channel Status Word (CSW) • I/O traffic controller
• cylinder • interblock gap (IBG)
• dedicated device • interrecord gap (IRG)
• direct access storage devices • interrupts
(DASDs) • LOOK
• direct memory access (DMA) • magnetic tape
• first come first served (FCFS) • optical disc drive
• I/O channel • polling
• I/O control unit • RAID
Terminology - 2
• rotational ordering • transport speed
• SCAN • virtual device
• search strategy
• search time
• seek strategy
• seek time
• sequential access media
• shared device
• shortest seek time first (SSTF)
• transfer rate
• transfer time