Chapter 6

External Memory
William Stallings
Computer Organization
and Architecture
10th Edition

1
 Types of External Memory
 Magnetic Disk
– HardDisk
– RAID (Redundant Array of Independent Disks)
– Removable (Floppy Disk, Zip)
 Optical
– CD-ROM
– CD-Recordable (CD-R)
– CD-R/W
– DVD
– Blu-Ray
– HD-DVD
 Magnetic Tape
 Flash Drive

2
 Magnetic Disk

A disk is a circular platter

 constructed of nonmagnetic material called the substrate
 Disk substrate coated with magnetizable material (iron oxide)

 Substrate used to be aluminium/aluminium-alloy

– Now glass

3
 Magnetic Disk

 Glass Substrate - benefits

– Improved surface uniformity
 Increases disk reliability
– Reduction in surface defects
 Reduced read/write errors
– Ability to support lower flight heights (See later)
– Better stiffness to reduce disk dynamics
– Better shock/damage resistance

4
Inductive Write MR Read

7
 Data Organization and Formatting

 Concentric rings or tracks

– Gaps between tracks
– Reduce gap to increase capacity
– Same number of bits per track (variable packing density)
– Constant angular velocity
 Tracksdivided into sectors
 Minimum block size is one sector
 May have more than one sector per block

9
Disk Data Layout - Tracks

10
 Disk Data Layout - Tracks
Sectors
Inter Track
Gaps

Tracks

11
 Disk Velocity
 Bit near centre of rotating disk passes fixed point slower than bit on outside of disk
 Increase spacing between bits in different tracks
 Rotate disk at constant angular velocity (CAV)
– Gives pie shaped sectors and concentric tracks
– Individual tracks and sectors addressable
– Move head to given track and wait for given sector
– Waste of space on outer tracks
 Lower data density
 Can use zones to increase capacity
– Each zone has fixed bits per track
– More complex circuitry

13
Disk Layout Methods Diagram

14
 Finding Sectors

 Mustbe able to identify start of track and sector

 Format disk
– Additional information not available to user
– Marks tracks and sectors

15
Winchester Disk Format
Seagate ST506

16
 Characteristics

1. Fixed or movable head

2. Removable or fixed
3. Single or double sided
4. Single or multiple platter
5. Head mechanism
– Contact (Floppy)
– Fixed gap
– Flying (Winchester)

17
 1. Fixed/Movable Head Disk

 Fixed head
– One read write head per track
– Heads mounted on fixed ridged arm

 Movable head
– One read write head per side
– Mounted on a movable arm

18
 2. Removable or Not

 Removable disk
– Can be removed from drive and replaced with another disk
– Provides unlimited storage capacity
– Easy data transfer between systems
 Nonremovable disk
– Permanently mounted in the drive

19
 3. Single or Double Sided

 Single Sided disk

– Data written only on one side of the disk

 Double sided disk

– Data written only on both sides of the disk

20
4. Multiple Platters

22
Tracks and Cylinders

24
Tracks and Cylinders

26
 5. Head mechanism

 Contact (Floppy)
 Fixed gap
 Flying (Winchester)
– Head must generate or sense an EM field of sufficient magnitude to write and
read properly.
– The narrower the head, the closer it must be to the platter
– Narrower head = narrower tracks = greater data density
– Closer the head, greater the risk of error from impurities/imperfections
Cont 

27
 5. Head mechanism
 Flying (Winchester Hard Disk)

 Head is an aerodynamic foil

 Air pressure generated by spinning disk is
enough to make the foil rise above the
surface
 Heads fly on boundary layer of air as disk
spins
 Very small head to disk gap

28
 Winchester Hard Disk (2)

 Universal
 Cheap
 Fastest external storage
 Getting larger all the time

30
 Disk Performance Parameters

 Actual details of disk I/O operation depend on

– The computer system
– The operating system
– Nature of the I/O channel and disk controller h/w

32
 Disk Performance Parameters

 Seek time
– Moving head to correct track
 Rotational delay (latency)
– Waiting for data to rotate under head
 Access time = Seek + Latency
– Time taken by the head to get into read/write position
 Transfer rate

33
Timing of Disk I/O Transfer

34
 Disk Performance Parameters

 Seek time
 Consists of
– Initial start up time
– Traversal time: Time taken to traverse the track
 Traversal time consists of
– Traversal time
– Settling time (of the head)
 Typical average seek time for contemporary hard disks is
10ms

35
 Disk Performance Parameters

 Rotational Delay
 With 15000 rpm hard disk, One revolution will take …
– 4ms
 Thus on average, the rotational delay = 2ms

 Floppy: 300 to 600 rpm

– Rotational delay = 100 to 50 ms

36
SOLID STATE DRIVES

38
 SOLID STATE DRIVES (SSD)

 The term solid state refers to electronic circuitry built with

semiconductors.
 An SSD is a memory device made with solid state components
that can be used as a replacement to a hard disk drive.
 The SSDs now on the market and coming on line use NAND
flash memory

39
 SSD Compared to HDD
 SSDs have the following advantages over HDDs:
 High-performance input/output operations per second (IOPS):
– Significantly increases performance I/O subsystems.
 Durability:
– Less susceptible to physical shock and vibration.
 Longer lifespan:
– SSDs are not susceptible to mechanical wear.
 Lower power consumption:
– SSDs use considerably less power than comparable-size HDDs.
 Quieter and cooler running capabilities:
– Less space required, lower energy costs, and a greener enterprise.
 Lower access times and latency rates:
– Over 10 times faster than the spinning disks in an HDD.

40
 Comparison of Solid State Drives and Disk
Drives
NAND Flash Drives Seagate Laptop Internal HDD
File copy/write speed 200–550 Mbps 50–120 Mbps
Less power draw, averages 2–3 More power draw, averages 6–
watts, resulting in 30+ minute 7 watts and therefore uses
Power draw/battery life battery boost more battery
Typically not larger than 512 Typically around 500 GB and 2
GB for notebook size drives, 1 TB max for notebook size
Storage capacity TB max for desktops drives; 4 TB max for desktops
Approx. $0.50 per GB for a 1- Approx. $0.15 per GB for a 4-TB
Cost TB drive drive

41
 SSD Organization
 Figure (next slide) illustrates a general view of the common architectural
system component associated with any SSD system.
 On the host system, the operating system invokes file system software to
access data on the disk.
 The file system, in turn, invokes I/O driver software.
 The I/O driver software provides host access to the particular SSD
product.
 The interface component in Figure (next slide) refers to the physical and
electrical interface between the host processor and the SSD peripheral
device.
 If the device is an internal hard drive, a common interface is PCIe. For
external devices, one common interface is USB

43
 Host System
SSD
Interface Operating System
Software
Controller
File System Software
Addressing
I/O driver Software

Data buffer/cache Error Correction

Interface

Flash Memory
Components

Flash Memory
Components

Flash Memory Solid State Drive Architecture

Components
44
 SSD Components
 In addition to the interface to the host system, the SSD contains the
following components:
 Controller:
– Provides SSD device level interfacing and firmware execution.
 Addressing:
– Logic that performs the selection function across the flash memory components.
 Data buffer/cache:
– High speed RAM memory components used for speed matching and to
increased data throughput.
 Error correction:
– Logic for error detection and correction.
 Flash memory components:
– Individual NAND flash chips
45
 SSD - Practical Issues

 Thereare two practical issues peculiar to SSDs that are not

faced by HDDs.
– First, SSD performance has a tendency to slow down as the device is
used.
– A second practical issue with flash memory drives is that a flash
memory becomes unusable after a certain number of writes.
– Continue

46
 SSD - Practical Issues
 First Issue:
 SSD performance has a tendency to slow down as the device is used.
– To understand the reason for this, you need to know that files are stored on
disk as a set of pages, typically 4 KB in length.
– These pages are not necessarily, and indeed not typically, stored as a
contiguous set of pages on the disk.
 (The reason for this arrangement is explained in the discussion of virtual memory in OS
Course.)
– However, flash memory is accessed in blocks, with a typical block size of 512
KB, so that there are typically 128 pages per block.
– Now consider what must be done to write a page onto a flash memory.
– Continue

47
 SSD - Practical Issues

1. The entire block must be read from the flash memory and
placed in a RAM buffer. Then the appropriate page in the
RAM buffer is updated.
2. Before the block can be written back to flash memory, the
entire block of flash memory must be erased—it is not
possible to erase just one page of the flash memory.
3. The entire block from the buffer is now written back to the
flash memory
Continue

48
 SSD - Practical Issues
 Now, when a flash drive is relatively empty and a new file is created, the pages of that file
are written on to the drive contiguously, so that one or only a few blocks are affected.
 However, over time, because of the way virtual memory works, files become fragmented,
with pages scattered over multiple blocks.
 As the drive become more occupied, there is more fragmentation, so the writing of a new
file can affect multiple blocks.
 Thus, the writing of multiple pages from one block becomes slower, the more fully
occupied the disk is.

 Manufacturers have developed a variety of techniques to compensate for this property of

flash memory, such as
– setting aside a substantial portion of the SSD as extra space for write operations (called
overprovisioning), then to erase inactive pages during idle time used to defragment the disk.
– Another technique is the TRIM command, which allows an operating system to inform an SSD
which blocks of data are no longer considered in use and can be wiped internally.

49
 SSD - Practical Issues
 A second practical issue with flash memory drives is that a flash memory becomes
unusable after a certain number of writes.

 As flash cells are stressed, they lose their ability to record and retain values.
 A typical limit is 100,000 writes.
 Techniques for prolonging the life of an SSD drive include
– front- ending the flash with a cache to delay and group write operations,
– using wear- leveling algorithms that evenly distribute writes across block of cells, and
– sophisticated bad-block management techniques.
 In addition, vendors are deploying SSDs in RAID configurations to further reduce the
probability of data loss.
 Most flash devices are also capable of estimating their own remaining lifetimes so systems
can anticipate failure and take preemptive action.

50