Chapter 12: Physical Storage Systems

Database System Concepts, 7th Ed.

©Silberschatz, Korth and Sudarshan
See www.db-book.com for conditions on re-use
 Classification of Physical Storage Media

 Can differentiate storage into:

 Volatile storage: loses contents when power is switched off
 Non-volatile storage:
 Contents persist even when power is switched off.
 Includes secondary and tertiary storage, as well as battery-
backed up main-memory.
 Can lose it contents – why?
 Factors affecting choice of storage media include
 Speed with which data can be accessed
 Cost per unit of data
 Reliability

Database System Concepts - 7th Edition 12.2 ©Silberschatz, Korth and Sudarshan
 Storage Hierarchy

Database System Concepts - 7th Edition 12.3 ©Silberschatz, Korth and Sudarshan
 Storage Terminology

 Primary storage: Fastest media but volatile (cache, main memory).

 Secondary storage: next level in hierarchy, non-volatile, moderately fast
access time
• Disks
• Also called on-line storage
 Tertiary storage: lowest level in hierarchy, non-volatile, slow access
time. Used primarily for archival storage
• Magnetic tape
 Sequential access, 1 to 12 TB capacity
 A few drives with many tapes
 Juke boxes with petabytes (1000’s of TB) of storage

• Also called off-line storage

Database System Concepts - 7th Edition 12.4 ©Silberschatz, Korth and Sudarshan
 Disks

 Two types of disks:

• Magnetic disk:
 Consists of several rotating platters magnetically coated
 Read-write head

• Solid state disks

 Data is stored on multiple flash storage devices
 Both types of disk consist of a number of disk block
 Disk block is a logical unit for storage allocation and retrieval
• Typically -- 4 to 16 kilobytes
• Smaller blocks: more transfers from disk
• Larger blocks: more space wasted due to partially filled blocks

Database System Concepts - 7th Edition 12.5 ©Silberschatz, Korth and Sudarshan
 Magnetic Disks

 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)
 Read-write head. 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.

Database System Concepts - 7th Edition 12.6 ©Silberschatz, Korth and Sudarshan
 Magnetic Hard Disk Mechanism

Schematic diagram of magnetic disk drive Photo of magnetic disk drive

Database System Concepts - 7th Edition 12.7 ©Silberschatz, Korth and Sudarshan
 Magnetic Disk Drives

 Platters range from .85” to 14” (historically)

• Commonly 3.5”, 2.5”, and 1.8”
 Range from 30GB to 3TB per drive
 Performance
• Transfer Rate – theoretical – 6 Gb/sec
• Effective Transfer Rate – real – 1Gb/sec
• Seek time from 3ms to 12ms – 9ms common for desktop drives
• Average seek time measured or calculated based on 1/3 of tracks
• Latency based on spindle speed
 1 / (RPM / 60) = 60 / RPM

• Average latency = ½ latency

Database System Concepts - 7th Edition 12.8 ©Silberschatz, Korth and Sudarshan
 Magnetic Disk Performance

 Access Latency = Average access time = average seek time + average

latency
• For fastest disk 3ms + 2ms = 5ms
• For slow disk 9ms + 5.56ms = 14.56ms
 Average I/O time = average access time + (amount to transfer / transfer
rate) + controller overhead
 For example to transfer a 4KB block on a 7200 RPM disk with a 5ms
average seek time, 1Gb/sec transfer rate with a .1ms controller overhead =
• 5ms + 4.17ms + 0.1ms + transfer time =
• Transfer time = 4KB / 1Gb/s * 8Gb / GB * 1GB / 10242KB = 32 /
(10242) = 0.031 ms
• Average I/O time for 4KB block = 9.27ms + .031ms = 9.301ms

Database System Concepts - 7th Edition 12.9 ©Silberschatz, Korth and Sudarshan
 The First Commercial Disk Drive

1956
IBM RAMDAC computer included
the IBM Model 350 disk storage
system

5M (7 bit) characters
50 x 24” platters
Access time = < 1 second

Database System Concepts - 7th Edition 12.10 ©Silberschatz, Korth and Sudarshan
 Storage Interfaces

 Disk interface standards families

• SATA (Serial ATA)
 SATA 3 supports data transfer speeds of up to 6 gigabits/sec

• SAS (Serial Attached SCSI)

 SAS Version 3 supports 12 gigabits/sec

• NVMe (Non-Volatile Memory Express) interface

 Works with PCIe connectors to support lower latency and
higher transfer rates
 Supports data transfer rates of up to 24 gigabits/sec
 Disks usually connected directly to computer system

Database System Concepts - 7th Edition 12.11 ©Silberschatz, Korth and Sudarshan
 Storage Attachment

 Computers access storage in three ways

• host-attached
• network-attached
• cloud
 Host attached access through local I/O ports, using one of several
technologies
• To attach many devices, use storage busses such as USB, firewire,
thunderbolt
• High-end systems use fibre channel (FC)
 High-speed serial architecture using fibre or copper cables
 Multiple hosts and storage devices can connect to the FC fabric

Database System Concepts - 7th Edition 12.12 ©Silberschatz, Korth and Sudarshan
 Network-Attached Storage

 Network-attached storage (NAS) is storage made available over a

network rather than over a local connection (such as a bus)
 Remotely attaching to file systems
 NFS and CIFS are common protocols
 Implemented via remote procedure calls (RPCs) between host and
storage over typically TCP or UDP on IP network
 iSCSI protocol uses IP network to carry the SCSI protocol
 Remotely attaching to devices (blocks)

Database System Concepts - 7th Edition 12.13 ©Silberschatz, Korth and Sudarshan
 Cloud Storage

 Similar to NAS, provides access to storage across a network

• Unlike NAS, accessed over the Internet or a WAN to remote data
center
 NAS presented as just another file system, while cloud storage is API
based, with programs using the APIs to provide access
• Examples include Dropbox, Amazon S3, Microsoft OneDrive,
Apple iCloud
• Use APIs because of latency and failure scenarios (NAS
protocols wouldn’t work well)

Database System Concepts - 7th Edition 12.14 ©Silberschatz, Korth and Sudarshan
 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 reading or writing the data
 Computes and attaches checksums to each sector to verify that
data is read back correctly
• 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 - 7th Edition 12.15 ©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.
 Average seek time is 1/2 the worst case seek time.
– Would be 1/3 if all tracks had the same number of sectors, and
we ignore the time to start and stop arm movement
 4 to 10 milliseconds on typical disks
 Rotational latency – time it takes for the sector to be accessed to
appear under the head.
 4 to 11 milliseconds on typical disks (5400 to 15000 r.p.m.)
 Average latency is 1/2 of the above latency.
 Overall latency is 5 to 20 msec depending on disk model
 Data-transfer rate – the rate at which data can be retrieved from or stored
to the disk.
 25 to 200 MB per second max rate, lower for inner tracks

Database System Concepts - 7th Edition 12.16 ©Silberschatz, Korth and Sudarshan
 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
• 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

Database System Concepts - 7th Edition 12.17 ©Silberschatz, Korth and Sudarshan
 Disk Access

 Sequential access pattern

• Successive requests are for successive disk blocks
• Disk seek required only for first block
 Random access pattern
• Successive requests are for blocks that can be anywhere on disk
• Each access requires a seek
• Transfer rates are low since a lot of time is wasted in seeks
 I/O operations per second (IOPS)
• Number of random block reads that a disk can support per second
• 50 to 200 IOPS on current generation magnetic disks

Database System Concepts - 7th Edition 12.18 ©Silberschatz, Korth and Sudarshan
 Solid State Disks (SSD)

 Use standard block-oriented disk interfaces, but store data on

multiple flash storage devices internally
 Transfer rate of up to 500 MB/sec using SATA, and
up to 3 GB/sec using NVMe PCIe

Database System Concepts - 7th Edition 12.19 ©Silberschatz, Korth and Sudarshan
 SSD Performance Metrics

 Random reads/writes per second

• Typical 4 KB reads: 10,000 reads per second (10,000 IOPS)
• Typical 4KB writes: 40,000 IOPS
• SSDs support parallel reads
 Typical 4KB reads:
• 100,000 IOPS with 32 requests in parallel (QD-32) on SATA
• 350,000 IOPS with QD-32 on NVMe PCIe
 Typical 4KB writes:
• 100,000 IOPS with QD-32, even higher on some models
 Data transfer rate for sequential reads/writes
• 400 MB/sec for SATA3, 2 to 3 GB/sec using NVMe PCIe
 Hybrid disks: combine small amount of flash cache with larger magnetic
disk

Database System Concepts - 7th Edition 12.20 ©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
 Mirroring (or shadowing)
 Duplicate every disk. Logical disk consists of two physical disks.
 Every write is carried out on both disks
 Reads can take place from either disk
 If one disk in a pair fails, data still available in the other
 Data loss will 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
 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 500*106 hours (or 57,000 years) for a
mirrored pair of disks (ignoring dependent failure modes)

Database System Concepts - 7th Edition 12.21 ©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
 Block-level striping

Database System Concepts - 7th Edition 12.22 ©Silberschatz, Korth and Sudarshan
 Improvement in Performance via Parallelism

 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
 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 - 7th Edition 12.23 ©Silberschatz, Korth and Sudarshan
 Redundant Arrays of Independ Disks (RAID)

 Disk organization techniques that manage a large numbers of disks,

providing a view of a single disk of
 High capacity and high speed by using multiple disks in parallel,
 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 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

Database System Concepts - 7th Edition 12.24 ©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 differing cost,
performance and reliability characteristics
 Various levels.
 Level 0
 Level 1
 Level 2 – not used in practice
 Level 3 – not used in practice
 Level 4 – not used in practice
 Level 5
 Level 6

Database System Concepts - 7th Edition 12.25 ©Silberschatz, Korth and Sudarshan
 RAID Level 0

 Block striping; non-redundant.

 Used in high-performance applications where data loss is not critical.
 Example with 4 disks
 Figure

 For a file with 20 blocks. Block 1 goes to first disk, block 2 goes to
second disk,

Database System Concepts - 7th Edition 12.26 ©Silberschatz, Korth and Sudarshan
 RAID Level 1

 Mirrored disks with block striping

 Offers best write performance.
 Popular for applications such as storing log files in a database system.
 Figure with 8 disks.
 4 with striping
 4 mirroring the other 4

Database System Concepts - 7th Edition 12.27 ©Silberschatz, Korth and Sudarshan
 RAID Level 5

 Block-interleaved distributed parity. Uses parity bits to enhance

redundancy at a lower cost
 Blocks are partitioned into sets, For a given set of blocks, a parity block is
computed and stored on a disk
 Parity block i stores XOR of bits from block i of each disk
 When writing data to a block i, parity block i must also be computed and
written to disk
 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)
 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
 To recover data for a block, compute XOR of bits from all other blocks in
the set including the parity block

Database System Concepts - 7th Edition 12.28 ©Silberschatz, Korth and Sudarshan
 RAID Level 5 (Cont.)

 Example 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.

 Block writes occur in parallel if the blocks and their parity blocks are on
different disks

Database System Concepts - 7th Edition 12.29 ©Silberschatz, Korth and Sudarshan
 RAID Level 5 (Cont.)

 Example, with an array of five disks

 The parity block, labeled Pk, for logical blocks k is stored in disk k mod 5
 The corresponding blocks of the other four disks store the four data blocks
4k to 4k+3.
 The following table indicates how the first 20 blocks, numbered 0 to 19, and
their parity blocks are laid out.

Database System Concepts - 7th Edition 12.30 ©Silberschatz, Korth and Sudarshan
 RAID Level 6

 P+Q Redundancy scheme

 Similar to Level 5, but stores two error correction blocks (P, Q)
instead of single parity block to guard against multiple disk failures.
 Better reliability than Level 5 at a higher cost
 Becoming more important as storage sizes increase
 Figure

Database System Concepts - 7th Edition 12.31 ©Silberschatz, Korth and Sudarshan
 RAID Levels Not Used in Practice

 RAID Level 2: Memory-Style Error-Correcting-Codes (ECC) with bit

striping.
 RAID Level 3: Bit-Interleaved Parity
 RAID Level 4: Block-Interleaved Parity; uses block-level striping, and
keeps a parity block on a separate parity disk for corresponding blocks
from N other disks.
 RAID 5 is better than RAID 4, since with RAID 4 with random writes,
parity disk gets much higher write load than other disks and becomes
a bottleneck

Database System Concepts - 7th Edition 12.32 ©Silberschatz, Korth and Sudarshan
 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
 Including time taken to rebuild failed disk

Database System Concepts - 7th Edition 12.33 ©Silberschatz, Korth and Sudarshan
 Choice of RAID Level

 RAID 0 is used only when data safety is not important

 E.g., data can be recovered quickly from other sources
 RAID 1 provides much better write performance than level 5
 Level 5 requires at least 2 block reads and 2 block writes to write a
single block, whereas Level 1 only requires 2 block writes
 RAID 1 has higher storage cost than level 5
 RAID 5 is preferred for applications where writes are sequential and large
(many blocks), and need large amounts of data storage
 RAID1 is preferred for applications with many random/small updates
 RIAD 6 gives better data protection than RAID 5 since it can tolerate two
disk (or disk block) failures
 Increasing in importance since latent block failures on one disk,
coupled with a failure of another disk can result in data loss with RAID
1 and RAID 5.

Database System Concepts - 7th Edition 12.34 ©Silberschatz, Korth and Sudarshan
 Optimization of Disk-Block Access

 Buffering: in-memory buffer to cache disk blocks

 Read-ahead: Read extra blocks from a track in anticipation that they
will be requested soon
 Disk-arm-scheduling algorithms re-order block requests so that disk
arm movement is minimized
 elevator algorithm

Database System Concepts - 7th Edition 12.35 ©Silberschatz, Korth and Sudarshan
 End of Chapter 12

Database System Concepts, 7th Ed.

©Silberschatz, Korth and Sudarshan
See www.db-book.com for conditions on re-use
 RAID Level 5

 RAID level 5 uses parity bits to enhance redundancy at a lower cost

 Blocks in a RAID system are partitioned into sets, For a given set of blocks,
a parity block is computed and stored on disk;
 The ith bit of the parity block is computed as the exclusive or (XOR) of
the ith bits of the all blocks in the set.
 If the contents of any one of the blocks in a set is lost due to a failure, the
block contents can be recovered by computing the bitwise-XOR of the
remaining blocks in the set, along with the parity block.
 Whenever a block is written, the parity block for its set must be recomputed
and written to disk. The new value of the parity block can be computed by
either
 (i) reading all the other blocks in the set from disk and computing the
new parity block, or
 (ii) by computing the XOR of the old value of the parity block with the old
and new value of the updated block.

Database System Concepts - 7th Edition 12.37 ©Silberschatz, Korth and Sudarshan
 RAID Level 5

 Parity blocks: Parity block i stores XOR of bits from block i of each disk
 When writing data to a block i, parity block i must also be computed
and written to disk
 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)
 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
 To recover data for a block, compute XOR of bits from all other blocks
in the set including the parity block

Database System Concepts - 7th Edition 12.38 ©Silberschatz, Korth and Sudarshan