Chapter 7, Storage

Discovering Computers 2004

Modified 11 May 04 1524 hrs.

Topics

Why We Want Storage Devices

Remember stuff we are not using at the moment.

Why We Want Removable Storage Devices
Archive. Store for permanent reference. 
Audit (compliance certification, financial audit: Whitehouse Travelgate.)
History.
Backup. Store copy of data off-site for disaster recovery. Store duplicate for use in data recovery.
Security. Store sensitive or proprietary data in vault when not in use.
Volume. Store more data than main memory and installed fixed disks can hold.
Mobility. Transferability. Duplicate for delivery to someone else using the same removable storage media.
Cost. Removable media are usually less expensive than non-removable media.

Properties of Storage Devices

Physical method of storage, recording, and retrieval: IBM http://www.almaden.ibm.com/st/disciplines/storage/ 
Electronic: semiconductor memory
Magnetic: magnetic disk, magnetic tape, magnetic card, magnetic core, magnetoelectronic memory
Topological: laser disk, CD, paper tape, punched card, IBM Millipede nanotech punch cards
Thermochemical: CD-R, CD-RW
Optical: holographic, recorded by 2 lasers, read with 1 laser. http://www.almaden.ibm.com/st/projects/holography/ 
Acoustic: Deltic (Delay Line Time Compression)
Data density
Lineal density, bits/unit distance : magnetic tape, paper tape 
Areal density, bits/unit area : disk, magnetic card 
IBM: 100 Gbits/in2 (= 12.5 GB/in2 ) or 15.5 Gbits/cm2 (= 1.94 GB/cm2 )for production model capacity of 400 GB drive.  May Wong, "IBM's New Disk Design Hold More Data", CompuServe Technology editorial, 21 May 2001.
6 GB/square inch at IBM: http://www.almaden.ibm.com/st/projects/patternedmedia/ . This is significantly higher density than a CD. (6 GB/in2 ) / (2.54 cm / in )2 = 930 MB / cm2. (Fall 2000)
IBM Millipede nanotech punch cards  1 Tbits/in2  (= 125 GB/in2 )   or   155 Gbits / cm2 
(= 19.4 GB / cm2 ). (11 JUN 2002)
Volume density, bits/unit volume: optical cube
Mass density, bits / kg: important for space vehicle and high performance aircraft applications.
Power density, bits / watt: important for battery operated volatile memory applications.
Capacity, number of bits (Digital Linear Tape is greatest)
Record speed, bits/s (semiconductor RAM is fastest)
Retrieval speed, bits/s (semiconductor RAM is fastest)
Volatility (semiconductor RAM, video storage tube, and deltic are volatile, others are nonvolatile)
Power requirement to sustain volatile memory (important for portable computers)
Power requirement to read or write nonvolatile memory
Cost of storage media per bit (tape is cheapest)
Shelf life, data decay, equipment obsolescence, access availability
Major problem in using only computer-based storage methods for data of long-term value.
NASA tapes of historic and expensive interplanetary satellite data are no longer readable.
Rerecording data to newer technology media requires a commitment of money, time, and people.  Punched card data can no longer be read. Not every valuable record has been rerecorded.
Cost of shelving has motivated short-sighted librarians to look to computer-based storage media to reduce archiving costs.
Data recorded on computer-based media requires people to be trained to use that media.
Controller Interface
USB
FireWire
EIDE (Fast ATA): Enhanced Integrated Drive Electronics, 4 disks at 137 GB/disk, transfer rate 66 MBps.
SCSI (Small Computer Systems Intervace): 8 or 15 devices, 160 MBps transfer rate

Excellent References

Hard disk, CD, CD-R, CD-RW, DVD, Tape: The PC Technology Guide http://www.pctechguide.com

Text

Primary storage: Main Memory.
volatile: 
Does not retain its state (does not remember) after power is removed.
semiconductor memory (most common in PCs today)
One approach to providing constant power to volatile memory logic is to use a capacitor that is periodically recharged (refreshed).  The rate of discharge is determined by the power requirement of the memory logic.  The refresh rate requirement is determined by the size of the capacitor used.
nonvolatile: 
Retains its state (remembers), even if power is removed.
magnetoelectronic memory  
(Honeywell begins production in 2001), magnetic core. The Feb 2000 issue of IEEE
Spectrum announced that Honeywell was scheduled to begin production in 2001
of magnetoelectronic memory. 
Stuart Parkin, "MagRAM: Non-Volatile Magnetic Random Access Memory", http://www.almaden.ibm.com/st/projects/magneto/mram/ 11 March 2002
Secondary storage: storage other than main memory
IBM Research on data storage http://www.almaden.ibm.com/st/disciplines/storage/ 11 MAR 2002
Access methods: 
sequential access: magnetic tape, magnetic bubble memory, mylar or aluminum tape, paper tape, punched cards, microfilm
Indexed sequential access (usually lumped with direct access): magnetic disk, magnetic drum, CD, CD-R, CD-RW, microfiche
direct (random) access: main memory, memory sticks, memory cards, associative memory
Storage medium
Reading, writing
Access time

Access time is the time between requesting data and receiving data.

When characterizing the performance of a storage device, access time is the average time, not the minimum time.  

When analyzing performance of a device controller scheduling algorithm (such as for disk or tape), access time is the time required to move from the current position to the new position, plus the data read time.

For a measure of access time for a single instance, the label of "minimum", "average", "maximum" do not apply.

Magnetic Disks

Magnetism  
Materials Research Society  http://www.mrs.org/ 22 MAR 2002
Magnetoelectronics = spin electronics = spintronics = use of electron spin to store information.
IBM http://www.almaden.ibm.com/st/disciplines/magnetism/ 11 MAR 2002
Characteristics of magnetic media
3.5-inch floppy: 
1.44 MB capacity, 2 sided, 80 tracks per side, 18 sectors per track, 512 bytes per sector
density: approximately 21 kB / cm2 or 134 kB / in2, based on 
720 kB / [ &pi ( (4.25 cm)2 - (2.65 cm)2 ) ]
Formatting: track, sector
Disk must be formatted before first use.
Most floppies are formatted at the factory. IBM format and MacIntosh format are not compatible. Look on the box before you buy.
You will lose access to data on a disk if you format the disk. Formatting a disk is one technique of "erasing" a disk to make it available for reuse. The FBI crime lab can possible retrieve some data from a disk that has subsequently been reformatted.
Use: file allocation table or directory, cluster
Windows records 2 File Allocation Tables on a floppy disk to improve reliability at the expense of record speed.
A cluster on a floppy drive consists of the two sectors.  A cluster on a hard disk consists of 1, 4, or 8 sectors, depending on the size of the logical drive. 
Write protection notch on floppy disk: Open to write protect.
Erasing: If you need to erase financial or privacy act data from a magnetic disk, you need to degauss the disk, not merely do a fast format.  A degaussing unit is an electromagnet.
Floppy disk drive: door or shutter, read/write head
The floppy disk read-write head rests on the floppy disk material while the disk is spinning.
The floppy disk spins at a much slower speed than a hard disk.
The floppy disk stops spinning after a short period of no data transfer activity.
Care of magnetic storage devices (including floppy disks)
Prevent exposure to unintended magnetic fields (permanent magnets, temporary magnets). Common home sources:  Refrigerator magnets (proud grandma display board), vacuum cleaner and fan electric motors.
Electric motors operate by a rotating magnetic field.
Resistive electrical appliances do not generate a significant magnetic field. Light bulbs, stove heating elements, hot water heating elements.
Don’t put floppy disks in glove compartment or trunk in the summer when these spaces get hot.
High capacity floppy disks: Zip Drive (96 MB [called 100 MB], 250 MB, 750 MB)
density: approximately 1.44 MB / cm2 or 9.3 MB / in2, based on 
50 MB / [ &pi ( (4.25 cm)2 - (2.65 cm)2 ) ]
96 MB Zip disk holds as much as 66 of the 1.44 MB floppy disks.  Iomega has discontinued the manufacture of 100 MB drives.
250 MB Zip disk holds as much as 173 of the 1.44 MB floppy disks.
750 MB Zip disk holds as much as 520 of the 1.44 MB floppy disks. The external USB drive transfers data at 7 MB/s.

Hard disk

Platters: mounted on the same spindle.
Capacity: 1.7 - 50 GB; function of density and area
FAT32 permits a 4 kB cluster size to accommodate addressing a 4 TB disk volume.
Fixed or removable disk
Fixed or movable heads
Directory 
location while disk is in use: some are copied into memory.
for single head system, place directory on tracks to minimize average head motion.
use dedicated head(s) (possibly fixed position heads) for directory tracks.
Optically assisted head positioning on advanced technology hard disks
Format: Defining the data storage locations.
partition: dividing a physical disk into more than one logical disk.
It is common on PCs for a hard drive to be partitioned into a C: and a D: drive.
low-level format: define starting boundaries of sectors on each track.
On large disks, such as used on main frame computers, the number of sectors per track can be increased for tracks further from the hub.  This makes better use of the surface area, increasing the capacity.
high-level format: define the number of sectors per cluster. Establish the file allocation table.
How a hard drive works
track mode, cylinder mode
disk caching
head crash
The hard disk is faster than the floppy disk because the hard disk 
spins continuously, 
spins faster [PC hard disks: 5400 or 7200 rpm], and 
has a higher density; for the same angular speed and radius, more bits per second pass under the read/write head. 
The floppy disk stops spinning after a few seconds of not being used.
The hard disk read-write head does not touch the disk during data operations.
Removable hard drives: Jaz disk (Iomega), disk pack
Hard disk controller
IDE, Integrated Drive Electronics: developed by Western Digital and Compaq in 1986. Can control a maximum of 2 standard hard drives (master, slave) . Typical sustained transfer rates are 1-3 MB/s, limited mostly by ISA bus. On lower performance disk drives, transfer rate is no more than 5 MB/s.
EIDE, Enhanced Integrated Drive Electronics: 1-2 hard disks, 14 GB capacity, 33 MB/s (compare to 2.4 MB/s for CD). ProCom makes a 16-bit EIDE controller that can handle 4 devices.  http://www.gy.com/spec/bfdmv.htm 
"Promise makes a good card, but I bought the ultra/66, and returned it for the ultra/33 . Got the /33 for 19.00 . And there is NO speed difference.  First off, you need an ultra/66 drive to take advantage of it, and most /66 drives out are 5,400 RPM. I bought an IBM ultra/33  7,200 RPM drive, which is faster, and cheaper."  http://discussions.hardwarecentral.com/Forum4/HTML/000538.html 
ATA-5 (1999): 66.6 MB/s maximum.
SCSI, Small Computer System Interface: 
parallel data transfer
costs more than EIDE
7 devices daisy chained, 100 MB/s
RAID: Redundant Array of Independent Disks
Mirroring, striping, combination
You can increase overall system speed by having more disks, and having sequential records of a file on different disks. This will permit simultaneous overlapping read and write operations, similar to the way you use interleaved memory.
Disk and file maintenance: Do maintenance monthly on a home system, and weekly on a small business system.
Check for bad sectors and prevent storage in those sectors. It is very difficult to make a disk with no bad sectors. A disk is usually made with excess capacity. The manufacturer blocks access to bad sectors when doing the low level formatting, and you still get the capacity advertised.  If a disk begins to fail due to aging or manufacturing defects, the rate of discovering failed sectors will increase exponentially over time.  If you see this starting to happen, back up your hard drive and replace it soon.
File Reorganization.
Defragmentation of disk.
On Windows systems, End Task for all programs except Explorer before running Defragmenter.  This greatly reduced time required to complete the process.
On large systems, file reorganization is needed when
sequential files are badly fragmented
direct access files have nearly filled the space allocated to them
when linked list file structures need to be rebalanced
Example Products, 28 DEC 2003
Quality ratings: Good = Seagate, IBM, Worst = Western Digital
Hitachi, 7200 rpm, 250 GB HD, 8.5 ms seek, 8 MB cache, 3.5 inch LP
SATA interface: $315
IDE interface: $224
Hitachi, 7200 rpm, 80 GB HD, 8.5 ms seek, 2 MB cache, 3.5 inch LP
IDE interface: $66
Western Digital, 10,000 rpm, 5.2 ms seek, 8 MB cache, 36.7 GB capacity
Serial ATA, 3.5 inch: $119
Hitachi, 15,000 rpm, 3.3 ms seek, 8 MB cache, 3.5 inch LP, 36.9 GB
SCSI 80 pin interface, $275
IBM, 15,000 rpm, 3.4 ms seek, 4 MB cache, 18.4 GB, 80 pin SCSI, $90

Laser Disk

Compact disk
Excellent discussion of CD, CD-R, CD-RW, Multi-Read by The PC Technology Guide, 24 NOV 2000. 05 JUN 2001: http://www.pctechguide.com/09cdr-rw.htm 
A good and brief technical review of digital audio CDs using engineering terminology and concepts: Michael C. McGoodwin, "Digital Audio CD and other Selected Digital Technologies
Based on Principles of Digital Audio, 4th Ed. by Ken C. Pohlmann and other sources", http://www.vgard.net/digitalaudio/digitalaudio.html 03 FEB 2002
Glossary of CD and DVD Technologies: http://www.cdpage.com/Compact_Disc_Glossary/glossarym.html 
FAQ on CD-R: http://www.cdrfaq.org/ 
Care and Feeding:  Store in a cool, dry, dark place.  UV lighting (including from fluorescent lamps) can harm a CD-R.
Laser burns pits into storage medium. Data integrity is not harmed by magnetic fields.
CD
Land (flat area) reflects light and is read as a one.
Pit absorbs light and is read as a zero.
CD-R
A CD-R writes data to a disc by using it's laser to physically burn pits into the organic dye. When heated beyond a critical temperature, the area ‘burned’ becomes opaque and reflects less light than areas that have not been heated by the laser.
The faster the writing speed the more susceptible a CD-R writer is to buffer underruns. 
Put a CD-R in a first generation DVD-ROM drive and it won’t reflect enough 650nm light for the drive to read the data. Subsequent, dual-wavelength head devices solved this problem. Also, some CD-ROM drives' lasers, especially older ones, may not be calibrated to read recordable CDs.
CD-RW
CD-RW commonly uses a crystalline compound. When it’s heated to one temperature and cooled it becomes crystalline, but if it’s heated to a higher temperature, when it cools down again it becomes amorphous. The crystalline areas allow the metalised layer to reflect the laser better while the non-crystalline portion absorbs the laser beam, so it is not reflected. 
'Direct overwriting' and the process can be repeated several thousand times per disc.
Single spiral track divided evenly into sectors, ISO 9660 format
CD disks read from the center out
Safeguard disks 
from scratches on the data side and from high temperatures.
You can clean it with a soft dry cloth. (Do not use solvents.) 
On the label side: You can glue paper to it. You can silk-screen it.  
Writing on the label side of a CD-ROM:  There is a technical issue about writing on the label side of a CD: ink can penetrate the plastic and damage the aluminum layer. This warning is made near the bottom of the Phillips site, http://www-eu.sv.philips.com/newtech/cd.html .

Write only on the premarked label side.

Use only an oil based felt marker. Do not use a hard-tip or ball-point pen.
Never write on the underside of a CD.

Do not expose to excessive heat or humidity.

Do not expose to direct sunlight.
Capacity: CD-ROM 670 MB, CD-RW 540 MB
Track (spiral) is 5.27km long: Adrian J Pullin, http://www.newi.ac.uk/pullina/ExtMem/tsld001.htm  
The total length of the helical 'track' of pits is about 5,800 m http://www.discusa.com/cdref/cdbasics/cdbasics3.htm 
Density: approximately 7 MB / cm2 or about 45 MB / in2 based on 
650 MB / [ &pi ((6 cm)2 - (2.5 cm)2 ) ]
Speed: 
Write speed / rewrite speed / read speed. Example: 40/10/40.  x1 = 150 kB/s.
Slow speed CDs use constant linear velocity (CLV) of 1.3 m/s (angular velocity varies)
32x and 40x data CD drives use constant angular velocity (CAV). The 40x CD has a transfer rate between 2.8 and 6 MB/s depending on where the data is located. (Disctronics Manufacturing Inc., http://www.discusa.com )
data transfer rate referenced to 150 kB/s. 16X, 40X, 75X
IBM 48X-20X(1) Black Internal CD-ROM Drive (31 March 2004): 3000 to 7200 KB/sec transfer rates and less than 85 ms access speed. 128KB buffer for look-ahead cache. http://www-132.ibm.com/webapp/wcs/stores/servlet/ProductDisplay?catalogId=-840&langId=-1&partNumber=10K3782&storeId=1 
Writing is slower than reading
Slower than a floppy disk. Good for archiving and for distribution. Not practical for frequently used general temporary secondary storage.
Single-session (record once only), multi-session (can record additional data); CD, CD-R, CD-RW
Open Session http://mcdbnt.lscf.ucsb.edu/mcdbcomputer/cdr/dcdread.htm
DVD (digital video disk): compared to CD
IEEE on DVD:  Alan E. Bell, "The dynamic digital disk", IEEE Spectrum, October,1999    Volume 36, Number 10, 
 http://www.spectrum.ieee.org/pubs/spectrum/9910/dvd.html 04 FEB 2002
DVD-ROM and CD-ROM store data using pits embossed in the disc surface. These pits are about four and one-half times as dense on the DVD-ROM as on the CD-ROM, yet store about seven times as much user-data per side.  The greater density is due to a more efficient data modulation scheme and error correction method that packs more user data bits into every pit.
DVD FAQ: http://www.dvddemystified.com/dvdfaq.html
higher density and capacity (4.7 - 15.9 GB)
Some CD-ROM drive manufacturers plan to cease CD-ROM drive production after a few years in favor of DVD-ROM drives.
two recording layers
some double sided
DVD-ROM
Pressed discs (the kind that movies come on) last longer than you will, anywhere from 50 to 300 years.
4.7 - 17 GB capacity
Sides Layers Capacity
1 1 4.7 GB
1 2 8.5 GB
2 1 9.4 GB
2 2 17 GB
DVD-R
DVD-R uses organic dye technology, like CD-R
First-generation capacity was 3.95 billion bytes, later extended to 4.7 billion bytes.
DVD-R and DVD+R discs are expected to last anywhere from 40 to 250 years, about as long as CD-R discs.
DVD-RW
DVD-RW (formerly DVD-R/W and also briefly known as DVD-ER) is a phase-change erasable format.
Capacity is 4.7 billion bytes. 
DVD-RW discs can be rewritten about 1,000 times.
The erasable formats (DVD-RAM, DVD-RW, and DVD+RW) are expected to last from 25 to 100 years.
DVD+RW
DVD+RW is an erasable format based on CD-RW technology.
The DVD+RW format uses phase-change media with a high-frequency wobbled groove that allows it to eliminate linking sectors.
1000 re-record cycles
3.0 GB / side
DVD+R
DVD+R is a write-once variation of DVD+RW, expected to appear in 2002. It's a dye-based medium, like DVD-R, so it will have similar compatibility as DVD-R.
DVD-RAM
DVD-RAM, with an initial storage capacity of 2.58 billion bytes, later increased to 4.7, uses phase-change dual (PD) technology with some magneto-optic (MO) features mixed in.
DVD-RAM can be rewritten more than 100,000 times, and the discs are expected to last at least 30 years.
4.7 GB / side
The tray that extends from a CD drive is not a cup holder! (No, it was not a sailor that tried it.)

Other Storage Devices

Magnetic Bubble Memory: Magnetic domains circulate around a loop. No moving parts. Nonvolatile. Good for extreme temperatures (hot and cold).  Intel made a unit in the early 1980s.

Comparison of DVD and CD-ROM. William L.R. Cruce, Ph.D., NE Ohio Universities College of Medicine
http://www.neoucom.edu/Cruce/Dvd98/introduction.html   
  DVD CD-ROM
Diameter (mm) 120 120
Disc Thickness (mm) 1.2 1.2
Substrate
Thickness (mm)		 0.6 1.2
Track Pitch (µm) 0.74 1.6
Minimum Pit
Size (µm)		 0.4 0.83
Wavelength (nm) 635/650 780
Single Layer
Capacity		 4.7 GB 650 MB

William Huttle, DVD Technology: http://www.nswc.navy.mil/cosip/nov97/cots1197-1.shtml (Nice technical overview)
Tape http://www.jegsworks.com/Lessons/lesson6/lesson6-8.htm 
Sequential access
Magnetic media (tapes and disks) lasts 10 to 30 years.
9-track (M4)
36-track cartridge (M490E)
Large capacity
Density 800 - 6250 bits/inch, 2400 ft (176 MB) - 4800 ft, 60 MB - 400 MB capacity http://webopedia.internet.com/TERM/t/tape.html 
Inexpensive medium
Some versions allow recording of digital and analog data simultaneously on different channels
Digital Audio Tape (DAT, DDS) 2 - 240 GB (DC 2004)
Digital Linear Tape (BASF) 70 GB capacity, 10 MB/s transfer rate http://www.transitionproducts.com/magmedia2.cfm#DTL 
20 - 220 GB (DC 2004)
Linear tape-open (LTO), 100 - 200 GB (DC 2004)
Magnetic tape streamers http://curriculum.calstatela.edu/faculty/dfrankl/kpe471/ch7/tsld034.htm 
cartridge used with microcomputer systems: 120 MB - 5 GB
4 mm, 2 GB - 24 GB capacity, with 180 kB/s transfer rate, developed for home audio http://www.9to5computer.com/MassStorage.asp and http://webopedia.internet.com/TERM/t/tape.html 
8 mm Helical Scan tape cartridges, 5 GB - 40 GB capacity at 500 kB/s transfer rate, developed for home video http://www.9to5computer.com/MassStorage.asp 
QIC (Quarter Inch Cartridge), 40 MB - 5 GB capacity http://webopedia.internet.com/TERM/t/tape.html 
40 MB - 25 GB (DC 2004)
Travan (TR), 8 - 40 GB (DC2004)
PC Cards
Other types of storage
Smart card: intelligent (process and store), memory (store only)
Military ID, Common Access Card (CAC)
Microfilm, microfiche
Archival-quality microfilm is projected to last 300 years or more.
Enterprise storage systems and data warehouses
Storage area network
Data warehouse
Punched Card (dimpled, pregnant, hanging chad, and all that) [museum]
IBM Millipede nanotech punch cards, possible future substitute for flash cards, http://www.computerworld.com/printthis/2002/0,4814,71913,00.html 
Paper tape, mylar tape [museum]
Paper (BOOK)
High-quality, acid-neutral paper can last 100 years or longer

Device Performance

Storage Density comparisons
Device Density
IBM hard disk technology record 930 MB / cm2
Compact Disk 7 MB / cm2
250 MB Zip disk 3.6 MB /  cm2
100 MB Zip disk 1.44 MB / cm2
3.5-inch 1.44 MB floppy  21 kB / cm2
9-track magnetic tape 17.7 kB / cm2

Speed comparisons
Device Access times
RDRAM 1.25 ns
SRAM: Static RAM (not refreshed as often as DRAM) 5 - 15 ns
SDRAM 7.5 ns
DRAM: Dynamic RAM (must be refreshed) 50 - 70 ns
EPROM 55 - 250 ns
ROM  25 - 250 ns
Hard disk 6 - 12 ms
Erasable optical 19 - 100 ms
Floppy disk 84 ms
CD-ROM 80 - 800 ms
DAT tape @ 20 s
QIC tape @ 40 s
8 mm tape 40 - 500 s

ms = 10-3 s = millisecond μs = 10-6 s = microsecond ns = 10-9 s = nanosecond ps = 10-12 s = picosecond

Useful Websites

Integrated Drive Electronics (IDE): http://www.ora.com/reference/dictionary/terms/I/Integrated_Drive_Electronics.htm 

Storage Review:  http://www.storagereview.com