Physical Database Design (ch. 16 & ch. 3) Introduction

¢¡¤£¥¡§¦¨¡©¡¡¤£
"!$#¨%&%#('()*
+¢,¤-¥,§.¨,/01,2,¤30402-
98":$;¨

@¢A¤B¥A§C¨ADEFAGA¤HEIEGB
NM"O$P¨Q&JQP(R(ST
JKLM
VYZ[V\V¤]Z^Z\W
U¢V¤W¥V§Ẍ
cb"d$ë f&_fe(g(hi
_`ab
Inputs to Physical Design
• Normalized relations.
• Volume estimates.
• Attribute definitions.
• Data usage: entered, retrieved, deleted, updated.
• Response time requirements.
• Requirements for security, backup, recovery, retention,
integrity.
• DBMS characteristics.
Physical Design Decisions
• Specifying attribute data types.
• Modifying the logical design.
• Specifying the file organization.
• Choosing indexes.
2

j¢k¤l¥k§m¨knopkqk¤rosoql
xw"y$z¨{&t{z(|(}~
tuvw
¢€¤¥€§‚¨€ƒ„…€†€¤‡„ˆ„†
Œ"Ž$¨&‰(‘(’“
‰Š‹Œ
Data Volumes and Query Frequencies
• Data volumes: estimation of number of data
records in each entity
• Query frequencies: estimation of number of
queries per hour towards each entity
These two types of information are useful for
determining the storage requirements and
performance requirements, which are needed
to make physical design decisions.
An Example
200 75
PART
1000
70
SUPLLIER
50
(
MANUFACTURED
PART
400
o
40% 70%
140
(
PURCHASED
PART
700
40
60
40
SUPLLY
2500
M
N
80
Quotation
3

”¢•¤–¥•§—¨•˜š•›•¤œ›–
¡ ¢¡"£$¤¨¥&ž¥¤(¦(§¨
žŸ
©¢ª¤«¥ª§¬¨ª­®¯ª°ª¤±®²®°«
·"¸$¹¨º&³º¹(»(¼½
³´µ
Designing Fields
• Choosing data type -- Char(8), Date, etc.
• Coding, compression, encryption.
• Controlling data integrity.
– Default value.
– Range control.
– Null value control.
– Referential integrity.
An example of code look-up table
PRODUCT File
Product_No Description Finish …
B100 Chair C
B120 Desk A
M128 Table C
T100 Bookcase B
… … …
FINISH Look-up Table
Code Value
A Birch
B Maple
C Oak
Coding is a way to achieve compression
4

¾¢¿¤À¥¿§Á¨¿ÂÃÄ¿Å¿¤ÆÃÇÃÅÀ
ÌË"Í$ΨÏ&ÈÏÎ(Ð(ÑÒ
ÈÉÊË
Ó¢Ô¤Õ¥Ô§Ö¨Ô×ØÙÔÚÔ¤ÛØÜØÚÕ
áà"â$ã¨ä&Ýäã(å(æç
ÝÞßà
Designing Fields
• Handling missing data.
– Substitute an estimate of the missing value.
– Trigger a report listing missing values.
– In programs, ignore missing data unless the
value is significant.
Physical Records
• Physical Record: A group of fields stored in
adjacent memory locations and retrieved
together as a unit.
• Page: The amount of data read or written in
one I/O operation. A page contains usually
a number of physical records.
• Blocking Factor: The number of physical
records per page.
5

è¢é¤ê¥é§ë¨éìíîéïé¤ðíñíïê
öõ"÷$ø¨ù&òùø(ú(ûü
òóôõ
¨þ£¢¥¤§¦þ©¨þ¤¤¨ÿ
ý¢þ¤ÿ¥þ¡
!"©#
£!
Designing Physical Files
• Physical File: A file as stored on the disk.
• Constructs to link two pieces of data:
– Sequential storage.
– Pointers.
• File Organization: How the files are
arranged on the disk.
• Access Method: How the data can be
retrieved based on the file organization.
Sequential File Organization
• Records of the file are stored in sequence by
the primary key field values.
Aces
Boilermakers
Devils
Flyers
Hawkeyes
Hoosiers
…
Minors
Panthers
…
Seminoles
…
Start of file
Scan
6

$&%'(%¡)%£*¥+§,-%©.¥%/+0+.1'
23456£5789298!:!;©<
=&>(>¡@>£A¥B§C->©D¥>EBFBD1
GHIJK£JLMNGNM!O!P©Q
Indexed File Organizations
• Index: an auxiliary file to improve access
efficiency of the main data file.
• B-tree or B + -tree index.
• Bitmap index
– Ideal for attributes that have even a few possible
values
– Often requires less storage space
– Can be used for multiple keys
Bitmap Index on Product Price attribute
Product Table Row Number
Price 1 2 3 4 5 6 7 8 9 10
100 0 0 1 0 1 0 0 0 0 0
200 1 0 0 0 0 0 0 0 0 0
300 0 1 0 0 0 0 1 0 0 1
400 0 0 0 1 0 1 0 1 1 0
Product 3 and 5 have Price $100
Product 1 has Price $200
Product 2, 7, and 10 have Price 300
Product 4, 6, 8, and 0 have Price $400
7

R&ST(S¡US£V¥W§X-S©Y¥SZW[WY1T
\]^_`£_abc\cb!d!e©f
g&hi(h¡jh£k¥l§m-h©n¥holpln1i
qrstu£tvwxqxw!y!z©{
Hashed File Organization
• Hashing Algorithm: Converts a primary key
value into a record address.
• Division-remainder method:
– Given 1000 pages to store employee records
– Choose the prime number closest to 1000, i.e., 997
– The bucket number of each record is equal to the
remainder of employee ID divided by 997
– Finding the location of any employee record needs
only a computation.
Comparison of File Organizations
• Sequential:
– No waste space
– Fast sequential retrieval
– no random retrieval
– update requires reorganization and slow
8

|&}~(}¡}£€¥§‚-}©ƒ¥}„…ƒ1~
†‡ˆ‰Š£‰‹Œ†Œ!Ž!©
‘&’“(’¡”’£•¥–§—-’©˜¥’–š–˜1“
¡¢›¢¡!£!¤©¥
›œžŸ£ž
Comparison of File Organizations
• Indexed
– require additional space for index
– support random retrieval
– deletion, addition, and update of records require
modification of indexes
Comparison of File Organizations
• Hashed
– May require overflow pages
– sequential retrieval is impractical
– random retrieval on primary key is very fast
since it does not need to access index
– deletion, addition, and modification of records
are relatively easy
9

¦&§¨(§¡©§£ª¥«§¬-§©­¥§®«¯«­1¨
°±²³´£³µ·°·!¸!¹©º
»&¼½(¼¡¾¼£¿¥À§Á-¼©Â¥¼ÃÀÄÀÂ1½
ÅÆÇÈÉ£ÈÊËÌÅÌË!Í!ΩÏ
Denormalization
• The reversal of normalization in order to
increase query processing efficiency.
• During physical database design,
denormalization is done if performance
consideration dominates the issue of
operational anomalies.
An example
• Two entities with a one-to-one relationship
SID
STUDET
1 1
Submit
SCHOLARSHIP
APPLICATION
AID
Address
STUDENT(SID, Address)
Application_
Date
Qualifications
APPLICATION(AID, Application_date, Qualification, SID)
Denormalized relation: STUDENT(SID, Address, AID, Application_Date,
Qualification)
10

Ð&ÑÒ(Ñ¡ÓÑ£Ô¥Õ§Ö-ѩץÑØÕÙÕ×1Ò
ÚÛÜÝÞ£ÝßàáÚáà!â!ã©ä
å&æç(æ¡èæ£é¥ê§ë-æ©ì¥æíêîêì1ç
ïðñòó£òôõöïöõ!÷!ø©ù
Partitioning
• Horizontal Partitioning: Distributing the
rows of a table into several separate files.
• Vertical Partitioning: Distributing the
columns of a table into several separate
files.
– The primary key must be repeated in each file.
Partitioning
• Advantages of Partitioning:
– Records used together are grouped together.
– Each partition can be optimized for performance.
– Security, recovery.
– Partitions stored on different disks.
– Take advantage of parallel processing capability.
• Disadvantages of Partitioning:
– Slow retrievals across partitions.
– Complexity.
11