Lecture Operating system principles - Chapter 11: I/O Management and disk scheduling

After studying this chapter, you should be able to: Summarize key categories of I/O devices on computers, discuss the organization of the I/O function, explain some of the key issues in the design of OS support for I/O, analyze the performance implications of various I/O buffering alternatives,...

Chapter 11 I/O Management and Disk Scheduling

– Operating System Design Issues

– I/O Buffering

– Disk Scheduling

– Disk Cache

• Solution: use a hierarchical modular design of I/O functions

– Hide details of device I/O in lower-level routines

– User processes and upper levels of OS see devices in terms of

general functions, such as read, write, open, close, lock, unlock

A Model of I/O Organization

• Logical I/O:

– Deals with the device as a logical resource

and is not concerned with the details of

actually controlling the device

– Allows user processes to deal with the device

in terms of a device identifier and simple

commands such as open, close, read, write

• Device I/O:

– Converts requested operations into sequence

of I/O instructions

A Model of I/O Organization

• Scheduling and Control:

– Performs actual queuing / scheduling and

control operations

– Handles interrupts and collects and

reports I/O status

– Interacts with the I/O module and hence

the device hardware

Goal: Efficiency

• Most I/O devices are extremely slow

compared to main memory

 I/O operations often form a bottleneck

in a computing system

• Multiprogramming allows some processes

to be waiting on I/O while another process

is executing

Goal: Efficiency

• Swapping brings in ready processes but

this is an I/O operation itself

• A major effort in I/O design has been

schemes for improving the efficiency of I/O

– I/O buffering

– Disk scheduling

– Disk cache

No Buffering

• Without a buffer, OS directly accesses the device as and when it needs

• A data area within the address space of

the user process is used for I/O

No Buffering

• Process must wait for I/O to complete

before proceeding

– busy waiting (like programmed I/O)

– process suspension on an interrupt (like

interrupt-driven I/O or DMA)

• Problems

– the program is hung up waiting for the

relatively slow I/O to complete

No Buffering

– It is impossible to swap the process out

completely because the data area must be

locked in main memory before I/O

– Otherwise, data may be lost or single-process deadlock may happen

• the suspended process is blocked waiting on the

I/O event, and the I/O operation is blocked waiting for the process to be swapped in

I/O Buffering

• It may be more efficient to perform input

transfers in advance of requests being

made and to perform output transfers

some time after the request is made

• Transfers are made a block at a time

• Can reference data by block number

– mouse and other pointing devices, and

– most other devices that are not secondary

storage

• Transfer information as a stream of bytes

Single Buffer

• OS assigns a buffer in the system portion

of main memory for an I/O request

Block Oriented Single Buffer

• Input transfers are made to system buffer

• Block moved to user space when needed

• The next block is immediately requested,

expecting that the block will eventually be

needed

– Read ahead or Anticipated Input

• A reasonable assumption as data is

usually accessed sequentially

Block Oriented Single Buffer

•  Provide a speedup

– User process can be processing one block of data while the next block is being read in

•  OS is able to swap the process out

because the I/O operation is taking place

in system memory

Block Oriented Single Buffer

•  Complicate the logic in OS

– OS must keep track of the assignment of

system buffers to user processes

•  Affect the swapping logic

– Consider both the I/O operation and swapping involve the same disk

• Does it make sense to swap the process out after the I/O operation finishes?

Stream-oriented Single Buffer

• Line-at-time or Byte-at-a-time

• Terminals often deal with one line at a time with

carriage return signaling the end of the line

– Also line printer

• Byte-at-a-time suits devices where a single

keystroke may be significant

– Also sensors and controllers

– Interaction between OS and user process follows the

Double Buffer

• Use two system buffers instead of one

• A process can transfer data to or from one buffer while OS empties or fills the other

buffer

Circular Buffer

• More than two buffers are used

• Each individual buffer is one unit in a circular

Disk Performance

Parameters

• Currently, disks are at least four orders of

magnitude slower than main memory

 performance of disk storage subsystem is of

vital concern

• A general timing diagram of disk I/O transfer is

shown here.

Positioning the Read/Write Heads

• When the disk drive is operating, the disk

is rotating at constant speed

• To read or write, the head must be

positioned at the desired track and at the

beginning of the desired sector on that

track

• Track selection involves moving the head

in a movable-head system

Disk Performance

Parameters

• Access Time is the sum of:

– Seek time: The time it takes to position the

head at the desired track

– Rotational delay or rotational latency: The

time it takes for the beginning of the sector to reach the head

• Transfer Time is the time taken to transfer

the data (as the sector moves under the

Disk Performance

Parameters

• Total average access time T a

T a = T s + 1 / (2r) + b / (rN)

where T s = average seek time

b = no of bytes to be transferred

N = no of bytes on a track

r = rotation speed, in revolutions / sec.

• Due to the seek time, the order in which

sectors are read from disk has a

tremendous effect on I/O performance

Disk Scheduling

Policies

• To compare various schemes, consider a

disk head is initially located at track 100

– assume a disk with 200 tracks and that the

disk request queue has random requests in it

• The requested tracks, in the order

received by the disk scheduler, are

– 55, 58, 39, 18, 90, 160, 150, 38, 184.

First-in, first-out (FIFO)

• Process requests sequentially

• Fair to all processes

• May have good performance if most requests

are to clustered file sectors

• Approaches random scheduling in performance

if there are many processes

disk arm movement

• Control of the scheduling is outside the

control of disk management software

• Goal is not to optimize disk use but to

meet other objectives

• Often, short batch jobs & interactive jobs

are given higher priority than longer jobs

– Provide high throughput & good interactive

response time

– Longer jobs may have to wait an excessively

Last-in, first-out

• Good for transaction processing systems

– The device is given to the most recent user so there should be little arm movement for

moving through a sequential file

• Possibility of starvation since a job may

never regain the head of the line

Shortest Service

Time First

• Select the disk I/O request that requires

the least movement of the disk arm from

its current position

• Always choose the minimum seek time

• Arm moves in one direction only, satisfying all outstanding requests until it reaches the last track in that direction then the

direction is reversed

• LOOK policy: reverse direction when there are no more requests in a direction

– jobs whose requests are for tracks

nearest to both innermost and

outermost tracks and

C-SCAN (Circular SCAN)

• Restricts scanning to one direction only

• When the last track has been visited in one

direction, the arm is returned to the opposite end

of the disk and the scan begins again

• Reduces the maximum delay experienced by

new requests

• With SSTF, SCAN, C-SCAN, the arm may not

move if processes monopolize the device by

repeated requests to one track: arm stickiness

• Segments the disk request queue into

• Two subqueues

• When a scan begins, all of the requests

are in one of the queues, with the other

empty

• During the scan, all new requests are put

into the other queue

• Service of new requests is deferred until all of the old requests have been processed

Performance Compared

Comparison of Disk Scheduling Algorithms

Disk Cache

• Buffer in main memory for disk sectors

• Contains a copy of some of the sectors

• When an I/O request is made for a

particular sector, a check is made to

determine if the sector is in the disk cache

– If so, the request is satisfied via the cache

– If not, the requested sector is read into the

Disk Cache

• Locality of reference

– When a block of data is fetched into the cache

to satisfy a single I/O request, it is likely that

there will be future references to that same

block

• One design issue: replacement strategy

– When a new sector is brought into the disk

cache, one of the existing blocks must be

Least Recently Used (LRU)

• The block that has been in the cache the

longest with no reference to it is replaced

• A stack of pointers reference the cache

– Most recently referenced block is on the top of the stack

– When a block is referenced or brought into the cache, it is placed on the top of the stack

– The block on the bottom of the stack is to be

LRU Disk Cache

Performance

• The miss ratio is, principally, a function of

the size of the disk cache

Least Frequently Used (LFU)

• The block that has experienced the fewest references is replaced

• A counter is associated with each block

– Incremented each time the block is accessed

• When replacement is required, the block

with the smallest count is selected

Consider certain blocks are referenced

repeatedly in short intervals due to locality, but

Refined Frequency-Based

Replacement

• Only blocks in the old section are eligible for

replacement

• Allows relatively frequently referenced blocks a

chance to build up their reference counts before becoming eligible for replacement

• Simulation studies indicate that this refined