Data Storage and Backup

Sanjay Goel
School of Business
University at Albany, SUNY
 Data Backup

Sanjay Goel, Information Technology Management Department, 2

School of Business, University at Albany, SUNY
Data Backup
Why?
• Files can be accidentally deleted
• Mission-critical data can become corrupt.
• Natural disasters can leave your office in ruin.
• Backup is the best insurance against disasters
– A backup is the most cost effective technique for
managing disasters
– You need to figure out the backup strategy that suits the
organization

Sanjay Goel, Information Technology Management Department, 3

School of Business, University at Albany, SUNY
Data Backup
Planning
• Planning involves
– Figuring out what data needs to be backed up
– How often the data should be backed up
• Boils down to risk analysis
• There are several pertinent questions that need to be answered
– How important is the data on your systems?
– What type of information does the data contain?
– How often does the data change?
– How quickly do you need to recover the data?
– Do you have the right equipment for backups?
– Who will be responsible for backup and recovery plan?
– What is the best time to schedule a backup?
– Do you need to store information off-site?
Sanjay Goel, Information Technology Management Department, 4
School of Business, University at Albany, SUNY
 Data Backup
Options
Type Description Pros Cons
Full A complete set Provides a complete Takes a long time and the most space
Backup of all files copy of all your data; on backup media; redundant backups
backed up. makes it easy to locate created, as most files remain static.
files for restoring.
Increm A backup of Uses the least time & Makes the job of restoring files difficult
ental files changed space as only files since the last full back u and
Backup since the last changed since the last subsequent incremental backups have
backup of any backup are copied; lets to be reinstalled in correct order. Also
type. you back up multiple makes it hard to locate a specific file in
versions of the same backup
file.
Differe A backup of Takes up less time and Redundant information stored, as each
ntial files changed space than a full backup stores much of the same
Backup since the last backup; provides for information plus information added
full backup. more efficient since the last full backup. Subsequent
restoration than differential backups take increasingly
incremental backups. longer as more files are changed. 5
Sanjay Goel, Information Technology Management Department,
Source: http://www.geekgirls.com
School of Business, University at Albany, SUNY
Data Backup
Types
• With differential backups all the files that have
changed since the last full backup are backed up
(which means that the size of the differential backup
grows over time).
• With incremental backups, only files that have
changed since the most recent full or incremental
backup are backed up (which means the size of the
incremental backup is usually much smaller than a
full backup).

Sanjay Goel, Information Technology Management Department, 6

School of Business, University at Albany, SUNY
Data Backup
Selecting Media and Devices
• Capacity: Amount of data that needs to be back up routinely
– Can the backup hardware support the required load
• Reliability: Failure rate of hardware and media
– Reliability needs to be balanced with cost and time.
• Extensibility: The extensibility of the backup solution.
– The solution needs to be scalable to growing needs of organization
• Speed: The speed at which data can be backed up & recovered.
– Need to balance cost of down operations versus cost of equipment
• Cost: The cost of the backup solution.
– Does it fit into your budget?
– Does it commensurate with the loss of data and services?

Sanjay Goel, Information Technology Management Department, 7

School of Business, University at Albany, SUNY
 Data Backup
Selecting Media and Devices
Device Media Capacity Speed Comments
3.5" Floppy 1.44 MB Nice for small amounts of data. Cheap and portable
Drive Removable Media Slow media.
Up to 700 MB Great backup device and wonderful for making your own
CD-R/W Removable Media Moderate music CDs too. Large backups will require multiple CDs.
Up to 4.7 GB Moderate
DVD-R/W Removable Media to Fast Similar to CD-R/W with greater storage space.
Hard Drive Up to 160 GB and Backup on current hard drive. Good for recovering files
(Primary) growing. Fast but insufficient against system failures.
Hard Drive up to 160 GB and New hard drives are cheap and somewhat easy to install.
(Alternate) growing. Fast External USB drives very efficient
This is a floppy on Steroids. The most popular high-
ZIP® Drive 100 MB or 250 MB Slow capacity floppy-disk type device.
A great high-capacity removable media. Generally used by
Tape Drive 4GB to 110 GB Fast more sophisticated users.
Internet Depends mostly on internet connection speed. No
Backup Unlimited storage Moderate devices to mess with. Data is off-site.
Printer Unlimited pagesGoel, Information
Sanjay Very SlowTechnology
A paperManagement
backup is some times very effective
Department, 8
School of Business, University at Albany, SUNY
Data Backup
On-site vs. Off-site
ON-SITE OFF-SITE
• Advantages: • Advantages
– Its under your control – Saves organization time
– You know what is happening – Service Level agreement ensures
with the media guaranteed backup
– You know it is working • Disadvantages
– Good for large data files – No control
• Disadvantages – You don't know what they do with
– Fire means data loss your data
– If you have an accident while – Large data stores can't backed-up
backing up data is lost – Network can be expensive
– Restoring backups requires
effort

Sanjay Goel, Information Technology Management Department, 9

School of Business, University at Albany, SUNY
 Data Backup
RAID

• RAID: Redundant Arrays of Inexpensive Disks

• Uses multiple hard drives to enhance I/O
performance, reliability, and capacity
• Metric of performance: Mean Time Between Failure
(MTBF)
• MTBF of the raid is the MTBF of a single drive
divided by the total number of drives used in the
RAID
– Fault-tolerance may be increased by adding redundant
disk arrays

Sanjay Goel, Information Technology Management Department, 10

School of Business, University at Albany, SUNY
 Data Backup
RAID
• RAID encompasses any basic concepts that attempt
to combine physical disk space for either reliability,
capacity, or performance
• There is both hardware and software implementation
of RAID
– Hardware RAID requires a RAID Controller
– Software RAID requires CPU power to run

Sanjay Goel, Information Technology Management Department, 11

School of Business, University at Albany, SUNY
Data Backup
RAID Cont’d.

• A basic disk may be transformed into a dynamic

disk using by creating volumes the Windows disk
management options
• A volume is a storage unit made from free space on
a disk
• Volumes can be formatted with a file system and
assigned a drive letter
• There are five available types of volumes for
dynamic disks: simple, spanned, mirror, striped, and
RAID-5

Sanjay Goel, Information Technology Management Department, 12

School of Business, University at Albany, SUNY
Data Backup
RAID Cont’d.
• Simple
– Uses free space on a single drive
– Not fault tolerant
• Spanned
– Uses free space on multiple drives (32 max)
– Not fault tolerant
• Striped (RAID-0)
– Data is allocated alternately and evenly across multiple
physical disks
– Not fault tolerant, cannot be mirrored

Sanjay Goel, Information Technology Management Department, 13

School of Business, University at Albany, SUNY
Data Backup
RAID Cont’d.
• Mirrored (RAID-1)
– All of the data is redundantly duplicated on two disk arrays
– Fault tolerant
• RAID-5
– Data is striped across an array of three or more disks along
with a value that can be used to reconstruct the array called
parity
– Failing disks can be reconstructed from parity and remaining
data
– Fault tolerant

Sanjay Goel, Information Technology Management Department, 14

School of Business, University at Albany, SUNY
Data Backup
Storage Area Network
• Storage Area Network (SAN): A SAN is a high level
network connecting servers and storage devices for
block level I/O.
– Supports disk mirroring, backup and restore, archival and
retrieval of archived data, data migration from one storage
device to another, and the sharing of data among different
servers in a network.
– SANs can incorporate sub networks with network-attached
storage (NAS) systems.
• NAS: Network Attached Storage is a hard disk set up with
its own network address rather than being attached to
a specific computer.

Sanjay Goel, Information Technology Management Department, 15

School of Business, University at Albany, SUNY
Data Backup
Hard Drive Protection
• Drive Fitness Test (DFT) uses a PC-based program that
accesses special hard drive microcode, enabling users to
monitor hard drive operation.
• Self Monitoring Analysis and Report Technology (SMART) is
a monitoring system for computer hard disks to detect and
report on various indicators of reliability, in the hope of
anticipating failures
– Useful for predictable failures when some failure modes, especially
mechanical wear and aging, happen gradually over time.
– Not very useful for unpredictable failures, such as an electronic
component burning out.

Sanjay Goel, Information Technology Management Department, 16

School of Business, University at Albany, SUNY
Data Backup
Hard Disk Specifications
• Capacity
– The whole capacity or capacity of one disk
• Rotate Speed
– The speed a disk rotates in RPMs
• Average Seek Time
– Measure of drive speed in multi-user environments where
read and write request are uncorrelated
– 10ms is common for hard drives
• Average Latency
– The time it takes for the head of a hard drive to meet the
correct sector of the drive once on the correct cylinder
– Faster rotation speeds equals lower latency
Sanjay Goel, Information Technology Management Department, 17
School of Business, University at Albany, SUNY
Data Backup
Hard Disk Specification Cont’d.
• Average Access Time
– Command Overhead Time + Seek Time + Latency
• Buffer Size (Cache)
– A small, fast memory holding recently accessed data for
quick access
– Cache processors are much faster than main processors and
stores information based on temporal and spatial locality
– Temporal and spatial locality refer to when data is accessed
and what data is in close proximity to that data

Sanjay Goel, Information Technology Management Department, 18

School of Business, University at Albany, SUNY
Data Backup
Hard Disk Specification Cont’d.
• Noise and Temperature
– Usually derived from the motor
– Lower temperatures mean happy and healthy
hard drives

Sanjay Goel, Information Technology Management Department, 19

School of Business, University at Albany, SUNY
 Disk Structure

Sanjay Goel, Information Technology Management Department, 20

School of Business, University at Albany, SUNY
Disk Structure
Physical Structure of Hard Disk
• Hard disks are made up of multiple platters spinning
on a spindle that are read and written to by
electromagnetic pins called heads
• Data on the platters are stored in circular bands and
heads can read and write to a single band called a band
track
• Sections within a track are called sectors

Sanjay Goel, Information Technology Management Department, 21

School of Business, University at Albany, SUNY
Disk Structure
Physical Structure of Hard Disk Cont’d.
• Platters rotate at a constant speed
• Tracks on the inside of the platter are moving faster
than those at the edges
• To compensate for this data is dense on the inward
tracks and sparse on the edges to create smooth read
times
• One side of one platter will always be reserved for
hardware track positioning information

Sanjay Goel, Information Technology Management Department, 22

School of Business, University at Albany, SUNY
Disk Structure
Logical Organization of Hard Disk
• A sector is typically 512 bytes
• A cluster is a space reserved for data and are typically
the same size as a sector
• Since many files are larger than 512 bytes, not all data
fits into a single sector
• When a file is written onto consecutive clusters, the
clusters are called contiguous
• Contiguous clusters are read faster than fragmented
clusters

Sanjay Goel, Information Technology Management Department, 23

School of Business, University at Albany, SUNY
Disk Structure
Logical Organization of Hard Disk
• Larger cluster sizes allow for less fragmentation, but
increases the potential for more unused space within
the clusters
• Reducing fragmentation decreases the amount of
memory used to store the location of used and unused
portions of the hard disk

Sanjay Goel, Information Technology Management Department, 24

School of Business, University at Albany, SUNY
 Disk Structure
Hard Disk Interfaces: IDE/ATA
• IDE refers to Integrated Device Electronics, but ATA,
Advance Technology Attachment, is the real industry
standard name
• Most PCs have two IDE controllers that may support
two devices for a maximum of four hard drives
• PATA or parallel ATA uses ribbon cables to connect
hard drives to the motherboard
• When two hard drives are connected to the same IDE
controller, one must be designated the master and one
the slave through jumper cables

Sanjay Goel, Information Technology Management Department, 25

School of Business, University at Albany, SUNY
Disk Structure
Hard Disk Interfaces: SATA

• Is replacing the more common PATA

• Replaces ribbon cables and master/slave designation
with more airflow friendly cables
• Differences in transfer rates are expected to hit
600MB/s in 2007 with SATA II
• Speed increases are held back by hard drive mechanic
speed and the use of PATA controllers

Sanjay Goel, Information Technology Management Department, 26

School of Business, University at Albany, SUNY
Disk Structure
Hard Disk Interfaces: DMA

• Direct Memory Access is a function of the memory

bus that allows for direct transfer between hard drives
and memory through the IDE controller ignoring
passage through the CPU
• Bus Mastering DMA ignores the IDE controller all
together and makes direct transfers between hard
disks and memory

Sanjay Goel, Information Technology Management Department, 27

School of Business, University at Albany, SUNY
Disk Structure
Hard Disk Interfaces: USB
• Universal Serial Bus is a hardware bus that is supported
on many motherboards
• Allows many devices to connect to the bus
• USB transfer rates are about 12 Mbits/s while USB 2.0
rates are 480 Mbits/s
• Firewire is a lesser known, but common alternative to
USB that is better than standard USB
• USB 2.0 leveled the disparities specifically by increasing
bandwidth amongst other things

Sanjay Goel, Information Technology Management Department, 28

School of Business, University at Albany, SUNY
Disk Structure
Hard Disk Interfaces: SCSI
• Small Computer System Interface that is less common
that ATA on PCs but more prominent on servers
• SCSI tends to be faster, more reliable, and more
expensive than ATA
• Another advantage is that is can handle seven devices
to ATA’s two

Sanjay Goel, Information Technology Management Department, 29

School of Business, University at Albany, SUNY
 Disk Structure
Primary Formatting of Hard Disk
• Before restoring data, hard disk logical structures
must be set up through low level formatting,
partitioning, then high level formatting
• The disk is divided into MBR, DBR, DIR, FAT, and
DATA

Sanjay Goel, Information Technology Management Department, 30

School of Business, University at Albany, SUNY
Disk Structure
Low Level Format (LLF)
• Functions
– Test hard disk media
– Partition tracks for hard disk
– Arrange sectors on track by interleave
– Assign sector IDs and finish setting sectors
– Test the hard disk surface for damaged sectors and mark
them “bad”
– Write certain ASCII to each sector
• Techniques
– CMOS, disk tools, and debug programs on older systems
– Hard disk manufacturers now provide tools to handle the
tasks
Sanjay Goel, Information Technology Management Department, 31
School of Business, University at Albany, SUNY
Disk Structure
High Level Format (HLF)
• After the low level format, logical drives are created
• Drives are commonly named after alphabet letters such as C: or
D:
• Trying to access the drives will currently result in a DISK
MEDIA ERROR because they are empty
• To use them a file system must be created
• A high level format of DOS logic disk can be initiated by the
“format” command

Sanjay Goel, Information Technology Management Department, 32

School of Business, University at Albany, SUNY
Disk Structure
High Level Format Cont’d.
• Functions
– Assign local serial numbers to sectors from cylinder
that assigned by each logical drive
– Establish DBR in basic partition and load 3 system
files of DOS if there is “/S” parameter in command
– Establish file allocation table (FAT) in each logical
disk
– Establish File Directory Table in each logical disk

Sanjay Goel, Information Technology Management Department, 33

School of Business, University at Albany, SUNY
Disk Structure
High Level Format cont’d.
• When using the “format” command:
– Activate the DOS partition by “Format C:/s”
– Format other logical disks by “Format [ d: ]”
– Note that formatting a disk will lose all information stored on
the disk
– For the using disk without adjusting the partition, also may
carry on the fast format command “Format C:/Q”

Sanjay Goel, Information Technology Management Department, 34

School of Business, University at Albany, SUNY
Disk Structure
High Level Format: Windows
• Explore windows will show you different partitions in
different colors
• Click right key in partition you wish to format
• Select the type of format you wish to execute
• Types range from format, fast format, complete format,
etc.

Sanjay Goel, Information Technology Management Department, 35

School of Business, University at Albany, SUNY
Disk Structure
High Level Format: Partition Magic
• The program will show you different partitions in
different colors
• Click right key in partition you wish to format
• Choose “format”
• Confirm and acknowledge that formatting will erase
any existing data on your hard drive

Sanjay Goel, Information Technology Management Department, 36

School of Business, University at Albany, SUNY
Disk Structure
Data Storage Region of Hard Disk

• Hard disks are divided into divided into MBR, DBR,

DIR, FAT, and DATA
• MBR is created by the partition software
• DBR, DIR, FAT, and DATA are created by the high
level format
• The file system writes in data by rewriting FAT, DIR,
and DATA areas

Sanjay Goel, Information Technology Management Department, 37

School of Business, University at Albany, SUNY
Disk Structure
MBR
• The first physical sector of the first hard drive (cylinder
0, head 0, sector 1)
• Each hard drive has an MBR but not every BIOS can
start the running OS from every drive
• MBR is then loaded to a fixed point in memory where it
loads the OS

Sanjay Goel, Information Technology Management Department, 38

School of Business, University at Albany, SUNY
Disk Structure
DBR
• DOS Boot Record (cylinder 0, column 1, sector 1)
• First sector an OS visits
• Contains a boot program and a BIOS Parameter Block
(BPB)
• Boot program determines if the first two files in the
root directory of this partition are the root files for the
OS

Sanjay Goel, Information Technology Management Department, 39

School of Business, University at Albany, SUNY
Disk Structure
FAT
• File Allocation Table
• File system for MS-DOS
• Nearly universal OS support
• Reading and writing slow relating to fragmentation
on creation and deletion
• The numbers after FAT indicate the number of
cluster bits (FAT12, FAT16, FAT32, etc.)

Sanjay Goel, Information Technology Management Department, 40

School of Business, University at Albany, SUNY
Disk Structure
FAT Cont’d.

• Directory or File Directory Table (FDT)

• The root sector after a backup FAT
• Records each start cell, files
• OS can locate files on the outset of FAT and
FAT

Sanjay Goel, Information Technology Management Department, 41

School of Business, University at Albany, SUNY
Disk Structure
Cont’d.

• Where files are stored

• Largest portion of hard disk space

Sanjay Goel, Information Technology Management Department, 42

School of Business, University at Albany, SUNY
Disk Structure
MBR

• Given no hardware damage, MBR recovery

is the first step in partition recovery
• MBR may be recovered using the “Fdisk”
command,Fixmbr from Microsoft, and other
similar programs

Sanjay Goel, Information Technology Management Department, 43

School of Business, University at Albany, SUNY
Disk Structure
Partition Recovery

• In minor cases, the partition can be restored

automatically
• In other cases, it must be built up manually
using tools such as Norton Utilities 8.0,
DiskMan, PartitionMagic, or Partition Table
Doctor

Sanjay Goel, Information Technology Management Department, 44

School of Business, University at Albany, SUNY
Disk Structure
DBR Recovery

• Partition OS cannot be booted if the DBR is

damaged
• Functions of DBR are different for FAT and
NFTS partitions
• Formatting will restore DBR, but not data
• Parition Table Doctor and WinHex are
examples of programs that can help recover
DBR

Sanjay Goel, Information Technology Management Department, 45

School of Business, University at Albany, SUNY
Disk Structure
FAT Recovery

• If FAT1 is damaged and FAT2 is not, FAT2

may be used to cover FAT1
• This includes finding the start sector of
FAT2 and finding the total length of the
FAT table using programs like DiskEdit and
WinHex

Sanjay Goel, Information Technology Management Department, 46

School of Business, University at Albany, SUNY
Summary
Acknowledgements

• I would like to thank Daren Pon for helping

prepare these slides for this presentation
• Materials for the lecture have been taken from
several sources including
– Helix and Knoppix web sites.
– Book on Data Recovery

Sanjay Goel, Information Technology Management Department, 47

School of Business, University at Albany, SUNY