Chapter 10: Storage and File Structure

The original presentation is infused with more information and slides

by Verena Kantere

Database System Concepts, 6th and 7th Ed.

©Silberschatz, Korth and Sudarshan
See www.db-book.com for conditions on re-use
 Chapter 10: Storage and File Structure

 Overview of Physical Storage Media

 Magnetic Disks
 RAID
 Storage Access
 File Organization
 Organization of Records in Files
 Data-Dictionary Storage

Database System Concepts - 6th and 7th Edition 10.2 ©Silberschatz, Korth and Sudarshan
 Classification of Physical Storage Media

 Speed with which data can be accessed

 Cost per unit of data
 Reliability
 data loss on power failure or system crash
 physical failure of the storage device
 Can differentiate storage into:
 volatile storage: loses contents when power is switched off
 non-volatile storage:
4 Contents persist even when power is switched off.
4 Includes secondary and tertiary storage, as well as
battery-backed up main-memory.

Database System Concepts - 6th and 7th Edition 10.3 ©Silberschatz, Korth and Sudarshan
 Physical Storage Media

 Cache – fastest and most costly form of storage; volatile; managed

by the computer system hardware.
 Main memory:
 fast access (10s to 100s of nanoseconds; 1 nanosecond = 10–9
seconds)
 generally too small (or too expensive) to store the entire
database
4 capacities of up to a few Gigabytes widely used currently
4 Capacities have gone up and per-byte costs have
decreased steadily and rapidly (roughly factor of 2 every 2
to 3 years)
 Volatile — contents of main memory are usually lost if a power
failure or system crash occurs.

Database System Concepts - 6th and 7th Edition 10.4 ©Silberschatz, Korth and Sudarshan
 Physical Storage Media (Cont.)
 Flash memory
 Data survives power failure
 Data can be written at a location only once, but location can be
erased and written to again
4 Can support only a limited number (10K – 1M) of write/erase
cycles.
4 Erasing of memory has to be done to an entire bank of
memory
 Reads are roughly as fast as main memory
 But writes are slow (few microseconds), erase is slower
 Widely used in embedded devices such as digital cameras,
phones, and USB keys

Database System Concepts - 6th and 7th Edition 10.5 ©Silberschatz, Korth and Sudarshan
 Physical Storage Media (Cont.)
 Magnetic-disk
 Data is stored on spinning disk, and read/written magnetically
 Primary medium for the long-term storage of data; typically stores
entire database.
 Data must be moved from disk to main memory for access, and written
back for storage
4 Much slower access than main memory (more on this later)
 direct-access – possible to read data on disk in any order, unlike
magnetic tape
 Capacities range usually to some TB
4 Much larger capacity and smaller cost/byte than main
memory/flash memory
4 Growing constantly and rapidly with technology improvements
(factor of 2 to 3 every 2 years)
 Survives power failures and system crashes
4 disk failure can destroy data, but is rare
Database System Concepts - 6th and 7th Edition 10.6 ©Silberschatz, Korth and Sudarshan
 Physical Storage Media (Cont.)
 Optical storage
 non-volatile, data is read optically from a spinning disk using a laser
 CD-ROM (640 MB) and DVD (4.7 to 17 GB) most popular forms
 Blu-ray disks: 25 GB to 128 GB
 Write-one, read-many (WORM) optical disks used for archival storage
(CD-R, DVD-R, DVD+R)
 Multiple write versions also available (CD-RW, DVD-RW, DVD+RW,
and DVD-RAM)
 Reads and writes are slower than with magnetic disk
 Juke-box systems, with large numbers of removable disks, a few
drives, and a mechanism for automatic loading/unloading of disks
available for storing large volumes of data

Database System Concepts - 6th and 7th Edition 10.7 ©Silberschatz, Korth and Sudarshan
 Physical Storage Media (Cont.)
 Tape storage
 non-volatile, used primarily for backup (to recover from disk failure),
and for archival data
 sequential-access – much slower than disk
 very high capacity (20 to 100 TB tapes available – in December
2020 IBM and Fujifilm announced a new tape with 580 TB capacity)
 tape can be removed from drive Þ storage costs much cheaper
than disk, but drives are expensive
 Tape jukeboxes available for storing massive amounts of data
4 hundreds of terabytes (1 terabyte = 109 bytes) to even multiple
petabytes (1 petabyte = 1012 bytes)

Database System Concepts - 6th and 7th Edition 10.8 ©Silberschatz, Korth and Sudarshan
 Tape Storage Example
CERN is still using tapes as the primary permanent storage:
Up to about 1.6B particle collisions per second inside the LHC
experiment's detectors (updated 2024).
The CERN Data Centre processes on average 1PB of data per day. The
LHC experiments produce over 45 PB of data per week, and an additional
hundreds of PB of data are produced per year from other (non-LHC)
experiments.
Magnetic tapes are used as the main long-term storage medium and
data from the archive is continuously migrated to newer technology,
higher density tapes.
The CERN storage system, EOS, was created for the extreme LHC
computing requirements. EOS instances at CERN are more than 2B. EOS
has expanded for other data storage needs beyond high-energy physics,
with AARNET, the Australian Academic and Research Network, and the
EU Joint Research Centre for Digital Earth and Reference Data adopting
it for their big-data systems.
https://information-technology.web.cern.ch/sites/information-
technology.web.cern.ch/files/CERNDataCentre_KeyInformation_June2020V1.pdf

Database System Concepts - 6th and 7th Edition 10.9 ©Silberschatz, Korth and Sudarshan
 CERN Data Center
 Underfloor space of about 1m in the Data Center for cables

CERN-CO-0307026-01Photo Copyright of CERN

Database System Concepts - 6th and 7th Edition 10.10 ©Silberschatz, Korth and Sudarshan
 CERN Data Center
 Grafana monitoring of CERN Data Center resources

https://monit-grafana-open.cern.ch/d/000000884/it-overview?orgId=16

 The custodial copy of all of CERN’s physics data is stored on magnetic tapes at
the CERN Data Centre, also called the WLCG Tier-0
 In the CERN’s report of November 2021: 32 244 tapes store about 380 PB

Database System Concepts - 6th and 7th Edition 10.11 ©Silberschatz, Korth and Sudarshan
 CERN New Data Center
 On 23 February 2024, a new data centre was inaugurated on CERN’s site:
 It spans more than 6000 square metres and including six rooms for IT
equipment with a cooling capacity of 2 MW each
 The centre will host CPU (central processing unit) servers for physics
data processing as well as a small amount of CPU servers and storage
capacity for business continuity and disaster recovery (for example,
when data is corrupted).
 CERN’s main data centre on the Meyrin site (Switzerland) will continue
to house the majority of the Organization’s data storage capacity.
 The data from these experiments is fed into the Worldwide LHC Computing
Grid (WLCG), a collaboration of around 170 data centres distributed across
more than 40 countries, with a storage capacity of about 3 exabytes and one
million CPU cores distributed across the network.

https://home.cern/news/news/computing/new-data-centre-cern

Database System Concepts - 6th and 7th Edition 10.12 ©Silberschatz, Korth and Sudarshan
 Storage Hierarchy
cache

main memory

flash memory

magnetic disk

optical disk

magnetic tapes

Database System Concepts - 6th and 7th Edition 10.13 ©Silberschatz, Korth and Sudarshan
 Storage Hierarchy (Cont.)

 primary storage: Fastest media but volatile (cache, main

memory).
 secondary storage: next level in hierarchy, non-volatile,
moderately fast access time
 also called on-line storage
 E.g. flash memory, magnetic disks
 tertiary storage: lowest level in hierarchy, non-volatile, slow
access time
 also called off-line storage
 E.g. magnetic tape, optical storage

Database System Concepts - 6th and 7th Edition 10.14 ©Silberschatz, Korth and Sudarshan
 Magnetic Hard Disk Mechanism
track t spindle

arm assembly
sector s

cylinder c read–write
head

platter
arm

rotation
NOTE: Diagram is schematic, and simplifies the structure of actual disk drives

Database System Concepts - 6th and 7th Edition 10.15 ©Silberschatz, Korth and Sudarshan
 Magnetic Disks
 Read-write head
 Positioned very close to the platter surface (almost touching it)
 Reads or writes magnetically encoded information
 Surface of platter divided into circular tracks
 Over 50K-100K tracks per platter on typical hard disks
 Each track is divided into sectors
 A sector is the smallest unit of data that can be read or written
 Sector size typically 512 bytes
 Typical sectors per track: 500 to 1000 (on inner tracks) to 1000 to 2000 (on
outer tracks)
 To read/write a sector
 disk arm swings to position head on right track
 platter spins continually; data is read/written as sector passes under head
 Head-disk assemblies
 multiple disk platters on a single spindle (1 to 5 usually)
 one head per platter, mounted on a common arm
 Cylinder i consists of ith track of all the platters
Database System Concepts - 6th and 7th Edition 10.16 ©Silberschatz, Korth and Sudarshan
 Magnetic Disks (Cont.)

 Earlier generation disks were susceptible to head-crashes

 Surface of earlier generation disks had metal-oxide coatings which would
disintegrate on head crash and damage all data on disk
 Current generation disks are less susceptible to such disastrous failures,
although individual sectors may get corrupted
 Disk controller – interfaces between the computer system and the disk
drive hardware.
 accepts high-level commands to read or write a sector
 initiates actions such as moving the disk arm to the right track and
actually reading or writing the data
 Computes and attaches checksums to each sector to verify that data is
read back correctly
4 If data is corrupted, with very high probability stored checksum won’t
match recomputed checksum
 Ensures successful writing by reading back sector after writing it
 Performs remapping of bad sectors

Database System Concepts - 6th and 7th Edition 10.17 ©Silberschatz, Korth and Sudarshan
 Disk Subsystem

 Multiple disks connected to a computer system through a controller

Controllers functionality (checksum, bad sector remapping) often carried

out by individual disks; reduces load on controller
 Disk interface standards families
 ATA (AT adaptor) range of standards
 SATA (Serial ATA)
 SCSI (Small Computer System Interconnect) range of standards
 SAS (Serial Attached SCSI)
 Several variants of each standard (different speeds and capabilities)

Database System Concepts - 6th and 7th Edition 10.18 ©Silberschatz, Korth and Sudarshan
 Disk Subsystem
 Disks usually connected directly to computer system
 In Storage Area Networks (SAN), a large number of disks are
connected by a high-speed network to a number of servers
 In Network Attached Storage (NAS) networked storage provides a
file system interface using networked file system protocol (TCP/IP),
instead of providing a disk system interface
 The difference is in how the data is accessed. A SAN accesses data
as blocks, while a NAS accesses data as files

Database System Concepts - 6th and 7th Edition 10.19 ©Silberschatz, Korth and Sudarshan
 Performance Measures of Disks
 Access time – the time it takes from when a read or write request is issued
to when data transfer begins. Consists of:
 Seek time – time it takes to reposition the arm over the correct track.
4 Average seek time is 1/2 the worst case seek time.
4 4 to 10 milliseconds on typical disks
 Rotational latency – time it takes for the sector to be accessed to
appear under the head.
4 Average latency is 1/2 of the worst case latency.
4 4 to 11 milliseconds on typical disks (5400 to 15000 r.p.m.)
 Data-transfer rate – the rate at which data can be retrieved from or stored
to the disk.
 25 to 100 MB per second max rate, lower for inner tracks
 Multiple disks may share a controller, so the rate that controller can
handle is also important
4 E.g. SATA: 150 MB/sec, SATA-II 3Gb (300 MB/sec)
4 Ultra 320 SCSI: 320 MB/s, SAS (3 to 6 Gb/sec)
4 Fiber Channel (FC2Gb or 4Gb): 256 to 512 MB/s

Database System Concepts - 6th and 7th Edition 10.20 ©Silberschatz, Korth and Sudarshan
 Performance Measures (Cont.)

 Mean time to failure (MTTF) – the average time the disk is

expected to run continuously without any failure.
 Typically 3 to 5 years
 Probability of failure of new disks is quite low, corresponding to a
“theoretical MTTF” of 500,000 to 1,200,000 hours for a new disk
(about 57 to 136 years)
4 E.g., an MTTF of 1,200,000 hours for a new disk means that
given 1000 relatively new disks, on an average one will fail
every 1200 hours
 MTTF decreases as disk ages
 Most disks have an expected life span of about 5 years, and
have significantly higher rates of failure once they become more
than a few years old.

Database System Concepts - 6th and 7th Edition 10.21 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access
 Block – a contiguous sequence of sectors from a single track
 data is transferred between disk and main memory in blocks
 sizes range from 512 bytes to several kilobytes
4 Smaller blocks: more transfers from disk
4 Larger blocks: more space wasted due to partially filled blocks
4 Typical block sizes today range from 4 to 16 kilobytes
 Disk-arm-scheduling algorithms order pending accesses to tracks so
that disk arm movement is minimized
 elevator algorithm:

R6 R3 R1 R5 R2 R4

Inner track Outer track

Database System Concepts - 6th and 7th Edition 10.22 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Some algorithms for disk scheduling

 First Come-First Serve (FCFS)
 Shortest Seek Time First (SSTF)
 Elevator (SCAN)
 Circular SCAN (C-SCAN)
 LOOK
 C-LOOK

Database System Concepts - 6th and 7th Edition 10.23 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Given the following queue -- 95, 180, 34, 119, 11, 123, 62, 64 with the Read-write head
initially at the track 50 and the tail track being at 199 we see different algorithms.

FCFS – first come first served

Database System Concepts - 6th and 7th Edition 10.24 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Given the following queue -- 95, 180, 34, 119, 11, 123, 62, 64 with the Read-write head
initially at the track 50 and the tail track being at 199 we see different algorithms.

236 tracks traversed

Starvation risk
SSTF - Shortest Seek Time First

Database System Concepts - 6th and 7th Edition 10.25 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Given the following queue -- 95, 180, 34, 119, 11, 123, 62, 64 with the Read-write head
initially at the track 50 and the tail track being at 199 we see different algorithms.

230 tracks traversed

SCAN – elevator algorithm

Database System Concepts - 6th and 7th Edition 10.26 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Given the following queue -- 95, 180, 34, 119, 11, 123, 62, 64 with the Read-write head
initially at the track 50 and the tail track being at 199 we see different algorithms.

187 tracks traversed

C-SCAN – circular scan

Database System Concepts - 6th and 7th Edition 10.27 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

Given the following queue -- 95, 180, 34, 119, 11, 123, 62, 64 with the Read-write head
initially at the track 50 and the tail track being at 199 we see different algorithms.

157 tracks traversed

C-LOOK

Database System Concepts - 6th and 7th Edition 10.28 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk Block Access (Cont.)

 File organization – optimize block access time by organizing the

blocks to correspond to how data will be accessed
 E.g. Store related information on the same or nearby cylinders.
 Files may get fragmented over time
4 E.g. if data is inserted to/deleted from the file
4 Or free blocks on disk are scattered, and newly created file
has its blocks scattered over the disk
4 Sequential access to a fragmented file results in increased
disk arm movement
 Some systems have utilities to defragment the file system, in
order to speed up file access

Database System Concepts - 6th and 7th Edition 10.29 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk Block Access (Cont.)

 Nonvolatile write buffers speed up disk writes by writing blocks to a non-volatile RAM
buffer immediately
 Non-volatile RAM: battery backed up RAM or flash memory
4 Even if power fails, the data is safe and will be written to disk when power returns
 Controller then writes to disk whenever the disk has no other requests or request has
been pending for some time
 Database operations that require data to be safely stored before continuing can
continue without waiting for data to be written to disk
 Writes can be reordered to minimize disk arm movement
 Log disk – a disk devoted to writing a sequential log of block updates
 Used exactly like nonvolatile RAM
4 Write to log disk is very fast since no seeks are required
4 No need for special hardware (NV-RAM)
 File systems typically reorder writes to disk to improve performance
 Journaling file systems write data in safe order to NV-RAM or log disk
 Reordering without journaling: risk of corruption of file system data

Database System Concepts - 6th and 7th Edition 10.30 ©Silberschatz, Korth and Sudarshan
 Flash Storage
 NOR flash vs NAND flash
 NAND flash
 used widely for storage, since it is much cheaper than NOR flash
 requires page-at-a-time read (page: 512 bytes to 4 KB)
 transfer rate around 20 MB/sec
 solid state disks: use multiple flash storage devices to provide higher
transfer rate of 100 to 200 MB/sec
 erase is very slow (1 to 2 millisecs)
4 erase block contains multiple pages
4 remapping of logical page addresses to physical page addresses
avoids waiting for erase
– translation table tracks mapping
» also stored in a label field of flash page
– remapping carried out by flash translation layer
4 after 100,000 to 1,000,000 erases, erase block becomes unreliable
and cannot be used
– wear leveling
Database System Concepts - 6th and 7th Edition 10.31 ©Silberschatz, Korth and Sudarshan
 RAID
 RAID: Redundant Arrays of Independent Disks
 disk organization techniques that manage a large numbers of disks,
providing a view of a single disk of
4 high capacity and high speed by using multiple disks in parallel,
4 high reliability by storing data redundantly, so that data can be
recovered even if a disk fails
 The chance that some disk out of a set of N disks will fail is much higher than
the chance that a specific single disk will fail.
 E.g., a system with 100 disks, each with MTTF of 100,000 hours (approx.
11.4 years), will have a system MTTF of 1000 hours (approx. 41 days)
 Techniques for using redundancy to avoid data loss are critical with large
numbers of disks
 Originally a cost-effective alternative to large, expensive disks
 I in RAID originally stood for ``inexpensive’’
 Today RAIDs are used for their higher reliability and bandwidth.
4 The “I” is interpreted as independent
Database System Concepts - 6th and 7th Edition 10.32 ©Silberschatz, Korth and Sudarshan
 Improvement of Reliability via Redundancy

 Redundancy – store extra information that can be used to rebuild

information lost in a disk failure
 E.g., Mirroring (or shadowing)
 Duplicate every disk. Logical disk consists of two physical disks.
 Every write is carried out on both disks
4 Reads can take place from either disk
 If one disk in a pair fails, data still available in the other
4 Data loss would occur only if a disk fails, and its mirror disk also
fails before the system is repaired
– Probability of combined event is very small
» Except for dependent failure modes such as fire or
building collapse or electrical power surges

Database System Concepts - 6th and 7th Edition 10.34 ©Silberschatz, Korth and Sudarshan
 Improvement of Reliability via Redundancy
 Mean time to data loss depends on mean time to failure,
and mean time to repair
 E.g. MTTF of 100,000 hours, mean time to repair of 10 hours gives
mean time to data loss of 100,0002/(2 ∗ 10) = 500*106 hours (or
57,000 years) for a mirrored pair of disks (ignoring dependent failure
modes)

Explanation:
• Every hour there is a 1/100K probability that one disk will fail. (By the way, the
probability that both disks will fail together is (1/100K)*(1/100K)).
• If there is a 10 hour mean time to repair and one of them fails we need to calculate
this probability that the other fails the next 10 hours: 1/100K * (1/100K+…+1/100K)
(10 times).
• Thus the probability of common failure in a period of 10 hours is 10/100K^2.
• Since this can happen for either the one or the other disk, this probability is
10/100K^2+10/100K^2 = 2*10/100K^2.
• This is certain (P = 1) in 1/p therefore in 100K^2/2*10.

Database System Concepts - 6th and 7th Edition 10.35 ©Silberschatz, Korth and Sudarshan
 Improvement in Performance via Parallelism

 Two main goals of parallelism in a disk system:

1. Load balance multiple small accesses to increase throughput
2. Parallelize large accesses to reduce response time
 Improve transfer rate by striping data across multiple disks.
 Bit-level striping – split the bits of each byte across multiple disks
 In an array of eight disks, write bit i of each byte to disk i.
 Each access can read data at eight times the rate of a single disk.
 But seek/access time worse than for a single disk
4 Bit level striping is not used much any more
 Block-level striping – with n disks, block i of a file goes to disk (i mod n) + 1
 Requests for different blocks can run in parallel if the blocks reside on
different disks
 A request for a long sequence of blocks can utilize all disks in parallel

Database System Concepts - 6th and 7th Edition 10.36 ©Silberschatz, Korth and Sudarshan
 RAID Levels
 Schemes to provide redundancy at lower cost by using disk
striping combined with parity bits
 Different RAID organizations, or RAID levels, have different
cost, performance and reliability characteristics
 RAID Level 0: Block striping; non-redundant.
 Used in high-performance applications where data loss is not critical.
 RAID Level 1: Mirrored disks with block striping
 Offers best write performance.
 Popular for applications such as storing log files in a database system.

Database System Concepts - 6th and 7th Edition 10.37 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
 RAID Level 2: Memory-Style Error-Correcting-Codes (ECC) with bit striping
 Hamming code: The number of disks used to store information is equal to
the logarithm of the number of disks that are protecting the data.
 RAID Level 3: Bit-Interleaved Parity
 a single parity bit is enough for error correction, not just detection, since we
know which disk has failed
 Byte-level striping and dedicated disk for parity
4 When writing data, corresponding parity bits must also be computed and
written to a parity bit disk
4 To recover data in a damaged disk, compute XOR of bits from other
disks (including parity bit disk)

Database System Concepts - 6th and 7th Edition 10.38 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
 RAID Level 3 (Cont.)
 Faster data transfer than with a single disk, but fewer I/Os per
second since every disk has to participate in every I/O.
 Subsumes Level 2 (provides all its benefits, at lower cost).
 RAID Level 4: Block-Interleaved Parity; uses block-level striping,
and keeps a parity block on a separate disk for corresponding
blocks from N other disks.
 When writing data block, corresponding block of parity bits must
also be computed and written to parity disk
 To find value of a damaged block, compute XOR of bits from
corresponding blocks (including parity block) from other disks.

Database System Concepts - 6th and 7th Edition 10.39 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
 RAID Level 4 (Cont.)
 Provides higher I/O rates for independent block reads than Level 3
4 block read goes to a single disk, so blocks stored on different disks
can be read in parallel
 Provides high transfer rates for reads of multiple blocks than no-striping
 Before writing a block, parity data must be computed
4 Can be done by using old parity block, old value of current block and
new value of current block (2 block reads + 2 block writes)
4 Or by recomputing the parity value using the new values of blocks
corresponding to the parity block
– More efficient for writing large amounts of data sequentially
 Parity block becomes a bottleneck for independent block writes since
every block write also writes to parity disk

Database System Concepts - 6th and 7th Edition 10.40 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
Small Writes
RAID 3 writes
New D1 data

D1 D2 D3 D4 P

⊕
3 reads and
2 writes
involving all D1 D2 D3 D4 P
the disks

RAID 4 small writes

New D1 data

D1 D2 D3 D4 P

⊕
2 reads and ⊕
2 writes
involving just
D1 D2 D3 D4 P
two disks

Database System Concepts -- 66thth and

System Concepts Edition
7th Edition 10.36
10.41 ©Silberschatz, Korth
©Silberschatz, Korth and
and Sudarshan
Sudarshan
 RAID Levels (Cont.)
 RAID Level 5: Block-Interleaved Distributed Parity; partitions data and
parity among all N + 1 disks, rather than storing data in N disks and
parity in 1 disk.
 E.g., with 5 disks, parity block for nth set of blocks is stored on disk
(n mod 5) + 1, with the data blocks stored on the other 4 disks.

P0 0 1 2 3
4 P1 5 6 7
8 9 P2 10 11
12 13 14 P3 15
16 17 18 19 P4
Database System Concepts - 6th and 7th Edition 10.42 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
RAID Levels (Cont.)
Distributing Parity Blocks
RAID 4 RAID 5

Can be done in parallel

1 2 3 4 P0 1 2 3 4 P0

5 6 7 8 P1 5 6 7 P1 8
Time

9 10 11 12 P2 9 10 P2 11 12

13 14 15 16 P3 13 P3 14 15 16

By distributing parity blocks to all disks, some small

writes can be performed in parallel

Database System Concepts - 6th and 7th Edition 10.43 ©Silberschatz, Korth and Sudarshan
Database System Concepts - 6th Edition 10.38 ©Silberschatz, Korth and Sudarshan
 RAID Levels (Cont.)
 RAID Level 5 (Cont.)
 Higher I/O rates than Level 4.
4 Block writes occur in parallel if the blocks and their parity
blocks are on different disks.
 Subsumes Level 4: provides same benefits, but avoids bottleneck
of parity disk.
 RAID Level 6: P+Q Redundancy scheme; similar to Level 5, but
stores extra redundant information to guard against multiple disk
failures.
 Better reliability than Level 5 at a higher cost; not used as widely.

Database System Concepts - 6th and 7th Edition 10.44 ©Silberschatz, Korth and Sudarshan
 Choice of RAID Level
 Factors in choosing RAID level
 Monetary cost
 Performance: Number of I/O operations per second, and
bandwidth during normal operation
 Performance during failure
 Performance during rebuild of failed disk
4 Including time taken to rebuild failed disk
 RAID 0 is used only when data safety is not important
E.g. data can be recovered quickly from other sources

 Level 2 and 4 never used since they are subsumed by 3 and 5
 Level 3 is not used anymore since byte-striping forces single block
reads to access all disks, wasting disk arm movement, which
block striping (level 5) avoids
 Level 6 is rarely used since levels 1 and 5 offer adequate safety
for most applications

Database System Concepts - 6th and 7th Edition 10.45 ©Silberschatz, Korth and Sudarshan
 Choice of RAID Level (Cont.)
 Level 1 provides much better write performance than level 5
 Level 5 requires 2 block reads and 2 block writes to write a single
block, whereas Level 1 only requires 2 block writes
 Level 1 preferred for high update environments such as log disks
 Level 1 had higher storage cost than level 5
 disk drive capacities increasing rapidly (50%/year) whereas disk
access times have decreased much less (x 3 in 10 years)
 I/O requirements have increased greatly, e.g. for Web servers
 When enough disks have been bought to satisfy required rate of
I/O, they often have spare storage capacity
4 so there is often no extra monetary cost for Level 1!
 Level 5 is preferred for applications with low update rate,
and large amounts of data
 Level 1 is preferred for all other applications

Database System Concepts - 6th and 7th Edition 10.46 ©Silberschatz, Korth and Sudarshan
 Hardware Issues

 Software RAID: RAID implementations done entirely in software, with no

special hardware support
 Hardware RAID: RAID implementations with special hardware
 Use non-volatile RAM to record writes that are being executed
 Beware: power failure during write can result in corrupted disk
4 E.g. failure after writing one block but before writing the second in a
mirrored system
4 Such corrupted data must be detected when power is restored
– Recovery from corruption is similar to recovery from failed disk
– NV-RAM helps to efficiently detect potentially corrupted blocks
» Otherwise all blocks of disk must be read and compared with
mirror/parity block

Database System Concepts - 6th and 7th Edition 10.47 ©Silberschatz, Korth and Sudarshan
 Hardware Issues (Cont.)
 Latent failures: data successfully written earlier gets damaged
 a single disk failure can lead to data loss if a sector in one of the other
disks has a latent failure that remains undetected
 Data scrubbing:
 continually scan for latent failures, and recover from copy/parity
 Hot swapping: disk replacement while system is running, without power down
 Supported by some hardware RAID systems
 reduces time to recovery, and improves availability greatly
 Many systems maintain spare disks which are kept online, and used as
replacements for failed disks immediately on detection of failure
 Reduces time to recovery greatly
 Many hardware RAID systems ensure that a single point of failure will not stop
the functioning of the system by using
 Redundant power supplies with battery backup
 Multiple controllers and multiple interconnections to guard against
controller/interconnection failures
Database System Concepts - 6th and 7th Edition 10.48 ©Silberschatz, Korth and Sudarshan
 File Organization, Record Organization
and Storage Access

Database System Concepts - 6th and 7th Edition 10.51 ©Silberschatz, Korth and Sudarshan
 File Organization

 The database is stored as a collection of files. Each file is a

sequence of records. A record is a sequence of fields.
 One approach:
 assume record size is fixed
 each file has records of one particular type only
 different files are used for different relations
This case is easiest to implement; will consider variable length
records later.

Database System Concepts - 6th and 7th Edition 10.52 ©Silberschatz, Korth and Sudarshan
 Fixed-Length Records
 Simple approach:
 Store record i starting from byte n * (i – 1), where n is the size of
each record.
 Record access is simple but records may cross blocks
4 Modification: do not allow records to cross block boundaries

record 0 10101 Srinivasan Comp. Sci. 65000

 Deletion of record i: record 1 12121 Wu Finance 90000
alternatives: record 2 15151 Mozart Music 40000
record 3 22222 Einstein Physics 95000
 move records i + 1, . . ., n record 4 32343 El Said History 60000
to i, . . . , n – 1 record 5 33456 Gold Physics 87000
record 6 45565 Katz Comp. Sci. 75000
 move record n to i
record 7 58583 Califieri History 62000
 do not move records, but record 8 76543 Singh Finance 80000
link all free records on a record 9 76766 Crick Biology 72000
record 10 83821 Brandt Comp. Sci. 92000
free list
record 11 98345 Kim Elec. Eng. 80000

Database System Concepts - 6th and 7th Edition 10.53 ©Silberschatz, Korth and Sudarshan
 Deleting record 3 and compacting
record 0 10101 Srinivasan Comp. Sci. 65000
record 1 12121 Wu Finance 90000
record 2 15151 Mozart Music 40000
record 4 32343 El Said History 60000
record 5 33456 Gold Physics 87000
record 6 45565 Katz Comp. Sci. 75000
record 7 58583 Califieri History 62000
record 8 76543 Singh Finance 80000
record 9 76766 Crick Biology 72000
record 10 83821 Brandt Comp. Sci. 92000
record 11 98345 Kim Elec. Eng. 80000

Database System Concepts - 6th and 7th Edition 10.54 ©Silberschatz, Korth and Sudarshan
 Deleting record 3 and moving last record
record 0 10101 Srinivasan Comp. Sci. 65000
record 1 12121 Wu Finance 90000
record 2 15151 Mozart Music 40000
record 11 98345 Kim Elec. Eng. 80000
record 4 32343 El Said History 60000
record 5 33456 Gold Physics 87000
record 6 45565 Katz Comp. Sci. 75000
record 7 58583 Califieri History 62000
record 8 76543 Singh Finance 80000
record 9 76766 Crick Biology 72000
record 10 83821 Brandt Comp. Sci. 92000

Database System Concepts - 6th and 7th Edition 10.55 ©Silberschatz, Korth and Sudarshan
 Free Lists
 Store the address of the
first deleted record in the header
record 0 10101 Srinivasan Comp. Sci. 65000
file header
record 1
 Use this first deleted record record 2 15151 Mozart Music 40000
to store the address of the record 3 22222 Einstein Physics 95000
second deleted record, and record 4
so on record 5 33456 Gold Physics 87000
record 6
 Can think of these stored record 7 58583 Califieri History 62000
addresses as pointers since record 8 76543 Singh Finance 80000
they “point” to the location record 9 76766 Crick Biology 72000
of a record record 10 83821 Brandt Comp. Sci. 92000
record 11 98345 Kim Elec. Eng. 80000
 Reuse space for normal
attributes of free records to
store pointers. (No pointers
stored in in-use records)

Database System Concepts - 6th and 7th Edition 10.56 ©Silberschatz, Korth and Sudarshan
 Variable-Length Records

 Variable-length records arise in database systems in several ways:

 Storage of multiple record types in a file.
 Record types that allow variable lengths for one or more fields such as
strings (varchar)
 Record types that allow repeating fields (used in some older data
models).
 Attributes are stored in order
 Variable length attributes represented by fixed size (offset, length), with
actual data stored after all fixed length attributes
 Null values represented by null-value bitmap

Database System Concepts - 6th and 7th Edition 10.57 ©Silberschatz, Korth and Sudarshan
 Variable-Length Records
The figure shows an instructor record, the first three attributes of which,
‘ID’, ‘name’, and ‘dept name’ are variable-length strings, and the
fourth attribute ‘salary’ is a fixed-sized number. We assume that the
offset and length values are stored in 2 bytes each, for a total of 4 bytes
per attribute. The salary attribute is assumed to be stored in 8 bytes, and
each string takes as many bytes as it has characters.

Null bitmap (stored in 1 byte)

0000

21, 5 26, 10 36, 10 65000 10101 Srinivasan Comp. Sci.

Bytes 0 4 8 12 20 21 26 36 45

The figure also illustrates the use of a null bitmap, which indicates which
attributes of the record have a null value. In this particular record, if the
salary were null, the fourth bit of the bitmap would be set to 1, and the
salary value stored in bytes 12 through 19 would be ignored.

Database System Concepts - 6th and 7th Edition 10.58 ©Silberschatz, Korth and Sudarshan
 SQL Server Record Storage

Database System Concepts - 6th and 7th Edition 10.59 ©Silberschatz, Korth and Sudarshan
 Variable-Length Records: Slotted Page Structure
Block Header Records

Size # Entries Free Space

Location

End of Free Space

 Slotted page header contains:
 number of record entries
 end of free space in the block
 location and size of each record
 Records can be moved around within a page to keep them contiguous
with no empty space between them; entry in the header must be updated.
 Pointers should not point directly to record — instead they should point to
the entry for the record in header.

Database System Concepts - 6th and 7th Edition 10.60 ©Silberschatz, Korth and Sudarshan
 Organization of Records in Files

 Heap – a record can be placed anywhere in the file where there

is space
 Sequential – store records in sequential order, based on the
value of the search key of each record
 Hashing – a hash function computed on some attribute of each
record; the result specifies in which block of the file the record
should be placed
 Records of each relation may be stored in a separate file. In a
multitable clustering file organization records of several
different relations can be stored in the same file
 Motivation: store related records on the same block to
minimize I/O

Database System Concepts - 6th and 7th Edition 10.61 ©Silberschatz, Korth and Sudarshan
 Sequential File Organization
 Suitable for applications that require sequential processing of
the entire file
 The records in the file are ordered by a search-key

10101 Srinivasan Comp. Sci. 65000

12121 Wu Finance 90000
15151 Mozart Music 40000
22222 Einstein Physics 95000
32343 El Said History 60000
33456 Gold Physics 87000
45565 Katz Comp. Sci. 75000
58583 Califieri History 62000
76543 Singh Finance 80000
76766 Crick Biology 72000
83821 Brandt Comp. Sci. 92000
98345 Kim Elec. Eng. 80000

Database System Concepts - 6th and 7th Edition 10.62 ©Silberschatz, Korth and Sudarshan
 Sequential File Organization (Cont.)
 Deletion – use pointer chains
 Insertion –locate the position where the record is to be inserted
 if there is free space insert there
 if no free space, insert the record in an overflow block
 In either case, pointer chain must be updated
 Need to reorganize the file 10101 Srinivasan Comp. Sci. 65000
12121 Wu Finance 90000
from time to time to restore
15151 Mozart Music 40000
sequential order 22222 Einstein Physics 95000
32343 El Said History 60000
33456 Gold Physics 87000
45565 Katz Comp. Sci. 75000
58583 Califieri History 62000
76543 Singh Finance 80000
76766 Crick Biology 72000
83821 Brandt Comp. Sci. 92000
98345 Kim Elec. Eng. 80000

32222 Verdi Music 48000

Database System Concepts - 6th and 7th Edition 10.63 ©Silberschatz, Korth and Sudarshan
 Multitable Clustering File Organization
Store several relations in one file using a multitable clustering
file organization

department

instructor

Comp. Sci. Taylor 100000

multitable clustering 45564 Katz 75000
of department and 10101 Srinivasan 65000
instructor
83821 Brandt 92000
Physics Watson 70000
33456 Gold 87000
Database System Concepts - 6th and 7th Edition 10.64 ©Silberschatz, Korth and Sudarshan
 Multitable Clustering File Organization (cont.)

 good for queries involving department instructor, and for queries

involving one single department and its instructors
 bad for queries involving only department
 results in variable size records
 Can add pointer chains to link records of a particular relation

Comp. Sci. Taylor 100000

45564 Katz 75000
10101 Srinivasan 65000
83821 Brandt 92000
Physics Watson 70000
33456 Gold 87000

Database System Concepts - 6th and 7th Edition 10.65 ©Silberschatz, Korth and Sudarshan
 Data Dictionary Storage
The Data dictionary (also called system catalog) stores
metadata; that is, data about data, such as
 Information about relations
 names of relations
 names, types and lengths of attributes of each relation
 names and definitions of views
 integrity constraints
 User and accounting information, including passwords
 Statistical and descriptive data
 number of tuples in each relation
 Physical file organization information
 How relation is stored (sequential/hash/…)
 Physical location of relation
 Information about indices (Chapter 11 – 6th edition, Chapter 14 – 7th edition)

Database System Concepts - 6th and 7th Edition 10.66 ©Silberschatz, Korth and Sudarshan
 Relational Representation of System Metadata
Relation_metadata A!ribute_metadata
 Relational relation_name relation_name
representation on number_of_a!ributes a!ribute_name
disk storage_organization domain_type
location position
 Specialized data
length
structures
designed for
Index_metadata
efficient access, in
memory index_name
relation_name
index_type
User_metadata
index_a!ributes
user_name
encrypted_password
group
View_metadata
view_name
definition

Database System Concepts - 6th and 7th Edition 10.67 ©Silberschatz, Korth and Sudarshan
 Storage Access

 A database file is partitioned into fixed-length storage units called

blocks. Blocks are units of both storage allocation and data
transfer.
 Database system seeks to minimize the number of block transfers
between the disk and memory. We can reduce the number of
disk accesses by keeping as many blocks as possible in main
memory.
 Buffer – portion of main memory available to store copies of disk
blocks.
 Buffer manager – subsystem responsible for allocating buffer
space in main memory.

Database System Concepts - 6th and 7th Edition 10.68 ©Silberschatz, Korth and Sudarshan
 Buffer Manager

 Programs call on the buffer manager when they need a block

from disk.
1. If the block is already in the buffer, buffer manager returns
the address of the block in main memory
2. If the block is not in the buffer, the buffer manager
1. Allocates space in the buffer for the block
1. Replacing (throwing out) some other block, if required,
to make space for the new block.
2. Replaced block written back to disk only if it was
modified since the most recent time that it was written
to/fetched from the disk.
2. Reads the block from the disk to the buffer, and returns
the address of the block in main memory to requester.

Database System Concepts - 6th and 7th Edition 10.69 ©Silberschatz, Korth and Sudarshan
 Buffer-Replacement Policies
 Most operating systems replace the block least recently used
(LRU strategy)
 Idea behind LRU – use past pattern of block references as a
predictor of future references
 Queries have well-defined access patterns (such as sequential
scans), and a database system can use the information in a user’s
query to predict future references
 LRU can be a bad strategy for certain access patterns involving
repeated scans of data
4 For example: when computing the join of 2 relations r and s
by a nested loops
for each tuple tr of r do
for each tuple ts of s do
if the tuples tr and ts match …
 Mixed strategy with hints on replacement strategy provided
by the query optimizer is preferable

Database System Concepts - 6th and 7th Edition 10.70 ©Silberschatz, Korth and Sudarshan
 Buffer-Replacement Policies (Cont.)
 Pinned block – memory block that is not allowed to be written
back to disk.
 Toss-immediate strategy – frees the space occupied by a block
as soon as the final tuple of that block has been processed
 Most recently used (MRU) strategy – system must pin the
block currently being processed. After the final tuple of that block
has been processed, the block is unpinned, and it becomes the
most recently used block.
 Buffer manager can use statistical information regarding the
probability that a request will reference a particular relation
 E.g., the data dictionary is frequently accessed. Heuristic:
keep data-dictionary blocks in main memory buffer

Database System Concepts - 6th and 7th Edition 10.71 ©Silberschatz, Korth and Sudarshan
 End of Chapter 10

Database System Concepts, 6th and 7th Ed.

©Silberschatz, Korth and Sudarshan
See www.db-book.com for conditions on re-use
 Figure 10.03
(a) RAID 0: nonredundant striping

C C C C

(b) RAID 1: mirrored disks

P P P

(c) RAID 2: memory-style error-correcting codes

(d) RAID 3: bit-interleaved parity

(e) RAID 4: block-interleaved parity

P P P P P

(f) RAID 5: block-interleaved distributed parity

P P P P
P P

(g) RAID 6: P + Q redundancy

Database System Concepts - 6th and 7th Edition 10.73 ©Silberschatz, Korth and Sudarshan
 Figure 10.18

Disk 1 Disk 2 Disk 3 Disk 4

B1 B2 B3 B4
P1 B5 B6 B7
B8 P2 B9 B10
…

…
Database System Concepts - 6th and 7th Edition 10.74 ©Silberschatz, Korth and Sudarshan
 Figure in-10.1

P0 0 1 2 3
4 P1 5 6 7
8 9 P2 10 11
12 13 14 P3 15
16 17 18 19 P4

Database System Concepts - 6th and 7th Edition 10.75 ©Silberschatz, Korth and Sudarshan