I/O Management and Disk Scheduling

Chapter 11

Categories of I/O Devices

Human readable Used to communicate with the user

Printers, Display terminals, Keyboard, Mouse

Machine readable Used to communicate with electronic

equipment Disk and tape drives, Sensors, Controllers

Communication Used to communicate with remote devices

Modems, Network interface cards

Differences in I/O Devices

Data rate May be differences of several orders of

magnitude between the data transfer rates

Application (the use to which the device is put) Disk used to store files requires file-

management software Disk used to store virtual memory pages

depends on the hardware and software to support it

Terminal used by system administrator may have a higher priority

Differences in I/O Devices

Complexity of control Control of a disk may be more complex than a

printer

Unit of transfer Data may be transferred as a stream of bytes

for a terminal or in larger blocks for a disk

Data representation Encoding schemes (character code, parity)

Error conditions The nature, reporting and responding to errors

differ widely

Techniques for Performing I/O

Programmed I/O Process issues I/O command and busy-

waits for the operation to complete

Interrupt-driven I/O I/O command is issued Processor continues executing

instructions I/O module sends an interrupt when

done

Techniques for Performing I/O

Direct Memory Access (DMA) Processor requests DMA module to

transfer a block of data DMA module controls exchange of

data between main memory and the I/O device

Processor interrupted only after entire block has been transferred

Evolution of the I/O Function

Processor directly controls a peripheral deviceController or I/O module is added Processor uses programmed I/O without

interrupts Processor does not need to handle details of

external devices

Controller or I/O module with interrupts Processor does not spend time waiting for an

I/O operation to be performed

Evolution of the I/O Function

Direct Memory Access I/O module given direct access to memory Processor involved at beginning and end only

I/O module is a separate processor CPU directs I/O processor to execute an I/O

program in main memory A sequence of I/O activities performed

without interruption

I/O processor has its own local memory It is a computer in its own right Minimal CPU involvement

Direct Memory Access

DMA module transfers a block of data to/from memory one word at a time over the system busDMA and CPU share the same system busDMA unit steals bus cycles (from the CPU) to transfer data (cycle stealing)The CPU instruction cycle is suspended to allow the current data word transferThis is not an interrupt (no context saves)The CPU just pauses for one bus cycle

Operating System Design Issues

Efficiency It is important because I/O is often

the bottleneck for computation Most I/O devices are extremely slow

compared to processor and main memory Use of multiprogramming allows for some

processes to be waiting on I/O while another process executes

Swapping is used to bring in additional Ready processes, but is itself an I/O operation

Operating System Design Issues

Generality Desirable to handle all I/O devices in a

uniform manner Diversity of I/O devices makes this

difficult Use hierarchical, modular approach Hide most of the details of device I/O in

lower-level routines so that processes and upper levels see devices in general terms such as read, write, open, close, lock, unlock

I/O Buffering

Reasons for buffering Processes must wait for I/O to

complete before proceeding Certain pages must remain (locked) in

main memory during I/O Interferes with swapping decisions of the

OS (the process cannot be swapped out) Use system buffers for I/O

Two Types of I/O Devices

Block-oriented Information is stored in fixed sized blocks Transfers are made one block at a time Used for disks and tapes

Stream-oriented Transfer information as a stream of bytes Used for terminals, printers, communication

ports, mouse, and most other devices that are not secondary storage

Single Buffer

Operating system assigns a buffer in main memory for an I/O requestBlock-oriented Input transfers made to system buffer Process moves block to user space when

needed Another block is requested to be transferred

into the buffer Called read ahead (or anticipated input):

assume the block will be needed

I/O Buffering

Single Buffer

Block-oriented User process can process one block of data

while next block is read in Swapping can occur since input is taking

place in system memory, not user memory Operating system keeps track of

assignment of system buffers to user processes

Similar considerations apply to block-oriented output also

Single Buffer

Stream-oriented Can be line-at-a-time or byte-at-a-

time Line-at-a-time

Input and output is one line at a time; buffer holds a single line

Byte-at-a-time Buffer is similar to that in

producer/consumer model

I/O Buffering

Double buffer Use two system buffers instead of one A process can transfer data to or from one

buffer while the operating system empties or fills the other buffer

Circular buffer More than two buffers are used Each individual buffer is one unit in a

circular buffer Used when I/O operation must keep up with

process

I/O Buffering

Utility of Buffering

Smooths out peaks in I/O demand Increases OS efficiency Improves performance of processes

However, no amount of buffering will allow I/O to keep pace with a process having a constant demand Eventually, the buffers will all fill up

and the process will have to wait

Disk Performance Parameters

To read or write, the disk head must be positioned at the desired track and at the beginning of the desired sectorSeek time time it takes to position the head at the

desired track

Rotational delay or rotational latency time its takes for the beginning of the

sector to reach the head

Timing of a Disk I/O Transfer

Disk Performance Parameters

Access time Sum of seek time and rotational delay The time it takes to get in position to read

or write

Data transfer occurs as the sector moves under the headTypical average seek time = 5 to 10 msAverage rotational delay (10,000 rpm) = 3ms

Disk Scheduling Policies

Reading a file from disk: Best performance when the file is stored in

adjacent tracks, occupying all sectors (sequential access)

Worst performance when file is stored in sectors distributed randomly over the disk (random access)

When I/O requests to a disk from many processes are queued up, how to schedule these requests for optimal performance?Random scheduling gives worst performance

Disk Scheduling Policies

First-in, first-out (FIFO) Process requests from queue in the

order in which they arrived Fair to all processes Good if only a few processes with

clustered accesses Approaches random scheduling in

performance if there are many processes

Disk Scheduling Policies

Priority Goal is not to optimize disk use but to

meet other objectives Short batch jobs and interactive jobs

may have higher priority Provide good interactive response

time Long jobs may wait excessively long

Disk Scheduling Policies

Last-in, first-out (LIFO) Most recent request is processed Good for transaction processing

systems The device is given to the most recent user The user may be reading a sequential file

(good locality), so there should be little arm movement

Possibility of starvation since a job may never regain the head of the line

Disk Scheduling Policies

Shortest Service Time First (SSTF) Select the disk I/O request that

requires the least movement of the disk arm from its current position

Always incurs the minimum seek time Better performance than FIFO

Disk Scheduling Policies

SCAN Arm moves in one direction only,

satisfying all outstanding requests until it reaches the last track or no more requests in that direction

Direction is then reversed No starvation Biased against area most recently

traversed: poor exploitation of locality

Disk Scheduling Policies

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

Disk Scheduling Policies

N-step-SCAN Segments the disk request queue into

subqueues of length N Subqueues are processed one at a time, using

SCAN New requests added to another queue when a

queue is being processed

FSCAN Uses two subqueues All requests are in one queue when scan begins New requests are put in other queue during scan

RAID

Redundant Array of Independent Disks (RAID) A set (array) of physical disk drives viewed by

the OS as a single logical device Data are distributed across the disks The disks operate independently and in

parallel, improving performance Multiple or even single I/O requests can be handled

in parallel Improves reliability (guards against disk

failure) by storing redundant parity information

RAID (cont’d)

Use RAID instead of a single large disk Large disks are expensive Large disks are less reliable

Consists of seven levels (0 to 6), each designating a different design

RAID 0 (non-redundant)

RAID 0

Data are striped across the disks The disk is divided into strips The disk strips are mapped to consecutive

logical strips in round robin fashion The corresponding strips from each disk

constitute a stripe If there are n disks, up to n strips can be

transferred in parallel. Provides High data transfer capacity (single request, large

data) High I/O request rate (multiple requests, small data)

RAID 1 (redundancy through mirroring)

•Data is duplicated (each logical strip mapped to two disks)•Recovery from failure is simple

RAID 2 (redundancy through Hamming code)

•Every I/O request requires access to all disks (parallel access)•Strips are very small (usually, byte or word)•Hamming code calculated across corresponding bits on each data disk and stored on multiple parity disks

RAID 3 (bit-interleaved parity)

•Similar to RAID 2, but requires only a single parity disk•Simple parity calculation (XOR)

RAID 4 (block-level parity)

•Separate I/O requests can be satisfied in parallel (independent access)•Strips are relatively large

RAID 5 (block-level distributed parity)

•Similar to RAID 4, but distributes parity strips across all disks

RAID 6 (dual redundancy)

•Uses two different parity calculations