Transcript Ch8(更新)

Chapter 8:
Advanced SQL
註 : 於11版為Chapter 7
楊立偉教授
台灣大學工管系
2015 Fall
1
Processing Multiple Tables–Joins

Join–a relational operation that causes two or more tables with a
common domain to be combined into a single table or view

Equi-join–a join in which the joining condition is based on

Natural join–an equi-join in which one of the duplicate columns

Outer join–a join in which rows that do not have matching

Union join–includes all columns from each table in the join, and
equality between values in the common columns; common columns
appear redundantly in the result table
is eliminated in the result table
values in common columns are nonetheless included in the result
table (as opposed to inner join, in which rows must have matching
values in order to appear in the result table)
an instance for each row of each table
The common columns in joined tables are usually the primary key of the
dominant table and the foreign key of the dependent table in 1:M relationships
Chapter 8
2
Figure 8-2
Visualization of different join types with results
returned in shaded area
Chapter 8
3
SELECT Order.*, Customer.*, Product.*
FROM Order
JOIN Customer ON Order.c_id=Customer.id
JOIN Product ON Order.p_id=Product.id
Customer
id Name Gender
1 張三
男
2 李四
女
Product
id Name
1 電腦
2 相機
Order
c_id p_id date
1 2 20090910
2 1 20091015
c_id p_id date
id Name
1 2 20090910 1 張三
2 1 20091015 2 李四
Gender
男
女
id Name
2 相機
1 電腦
Equi-join的結果
c_id p_id date
id Name
1 2 20090910 1 張三
2 1 20091015 2 李四
X
Gender
男
女
id Name
2 相機
1 電腦
X
Natural join的結果
Chapter 8
最原始, 由等號連結
其中必有部份欄位之值
完全相同 (Join條件)
將之剔除不顯示
4
Emp
no
1
2
3
4
5
Dept
no name mgr_no
1 會計部
3
2 工程部
4
name dept_no
張三
1
李四
2
王五
1
毛六
2
陳七
3
no
1
2
3
4
name dept_no no name mgr_no Equi-join的結果
張三
1
1 會計部
3
最原始, 由等號連結
李四
2
2 工程部
4
王五
1
1 會計部
3
毛六
2
2 工程部
4
SELECT Emp.*, Dept.*
FROM Emp
JOIN Dept ON Emp.dep_no=Dept.no
←注意這筆
no name dept_no no
1 張三
1
1
2 李四
2
2
3 王五
1
1
4 毛六
2
2
5 陳七
3 null
Chapter 8
name mgr_no
會計部
3
工程部
4
會計部
3
工程部
4
null
null
Left outer join的結果
Left : 以左邊為主
Outer : 不管是否有關聯到, 均列出
SELECT Emp.*, Dept.*
FROM Emp LEFT OUTER JOIN Dept
ON Emp.dep_no=Dept.no
5
Emp
no
1
2
3
4
5
Dept
no name mgr_no
1 會計部
3
2 工程部
4
name dept_no
張三
1
李四
2
王五
1
毛六
2
陳七
3
no
1
2
3
4
name dept_no no name mgr_no
張三
1
1 會計部
3
李四
2
2 工程部
4
王五
1
1 會計部
3
毛六
2
2 工程部
4
SELECT Emp.*, Dept.*
FROM Emp
JOIN Dept ON Emp.dep_no=Dept.no
←注意這筆
Left inner join的結果
Left : 以左邊為主
Inner : 有關聯到的才列出
→結果又等同Equi-join
SELECT Emp.*, Dept.*
FROM Emp LEFT INNER JOIN Dept 6
ON Emp.dep_no=Dept.no
預設就是inner
很少特別指定
Chapter 8
SELECT *
FROM Customer_TPE
SELECT *
FROM Customer_HKG
Customer_TPE
id Name Gender
1 張三
男
2 李四
女
Customer_HKG
id Name Gender
3 王五
女
4 毛六
男
id Name Gender
1 張三
男
2 李四
女
3 王五
女
4 毛六
男
Union-join的結果
垂直合併
SELECT *
FROM Customer_TPE
UNION
SELECT *
FROM Customer_HKG
Chapter 8
兩張表格必需聯集相容 Union Compatible
→兩張表格有相同之欄位，
且相對應之欄位有相同值域
合併後的結果必需符合表格特徵
→任兩筆完全相同紀錄的會被合併
7
Figure 8-1 Pine Valley Furniture Company Customer and Order
tables with pointers from customers to their orders (how Join works)
有10筆訂單
Chapter 8
有15個客戶
8
Natural Join Example

For each customer who placed an order, what is the
customer’s name and order number?
Join involves multiple tables in FROM clause
SELECT CUSTOMER_T.CUSTOMER_ID, CUSTOMER_NAME, ORDER_ID
FROM CUSTOMER_T NATURAL JOIN ORDER_T ON
CUSTOMER_T.CUSTOMER_ID = ORDER_T.CUSTOMER_ID;
ON clause performs the equality
check for common columns of the
two tables
Chapter 8
Note: from Fig. 1, you see
that only 10 Customers
have links with orders
 Only 10 rows will be
returned from this INNER
join
9
Outer Join Example

List the customer name, ID number, and order
number for all customers. Include customer
information even for customers that do have an order
SELECT CUSTOMER_T.CUSTOMER_ID, CUSTOMER_NAME, ORDER_ID
FROM CUSTOMER_T LEFT OUTER JOIN ORDER_T
ON CUSTOMER_T.CUSTOMER_ID = ORDER_T.CUSTOMER_ID;
LEFT OUTER JOIN syntax with
ON causes customer data to
appear even if there is no
corresponding order data
會回傳15筆
Chapter 8
10
Unlike INNER join, this will
include customer rows with
no matching order rows
Results
Chapter 8
11
Multiple Table Join Example

Assemble all information necessary to create an
invoice for order number 1006
Four tables involved in this join
SELECT CUSTOMER_T.CUSTOMER_ID, CUSTOMER_NAME,
CUSTOMER_ADDRESS, CITY, SATE, POSTAL_CODE,
ORDER_T.ORDER_ID, ORDER_DATE, QUANTITY,
PRODUCT_DESCRIPTION, STANDARD_PRICE,
(QUANTITY * UNIT_PRICE)
FROM CUSTOMER_T, ORDER_T, ORDER_LINE_T, PRODUCT_T
WHERE CUSTOMER_T.CUSTOMER_ID = ORDER_T.CUSTOMER_ID
AND ORDER_T.ORDER_ID = ORDER_LINE_T.ORDER_ID
AND ORDER_LINE_T.PRODUCT_ID = PRODUCT.PRODUCT_ID
AND ORDER_T.ORDER_ID = 1006;
Chapter 8
Each pair of tables requires an equality-check condition in the WHERE clause,
matching primary keys against foreign keys
12
Multiple Table Join Example
SELECT CUSTOMER_T.CUSTOMER_ID, CUSTOMER_NAME,
CUSTOMER_ADDRESS, CITY, SATE, POSTAL_CODE, ORDER_T.ORDER_ID,
ORDER_DATE, QUANTITY, PRODUCT_DESCRIPTION, STANDARD_PRICE,
(QUANTITY * UNIT_PRICE)
FROM CUSTOMER_T, ORDER_T, ORDER_LINE_T, PRODUCT_T
WHERE CUSTOMER_T.CUSTOMER_ID = ORDER_LINE.CUSTOMER_ID
AND ORDER_T.ORDER_ID = ORDER_LINE_T.ORDER_ID
AND ORDER_LINE_T.PRODUCT_ID = PRODUCT.PRODUCT_ID
AND ORDER_T.ORDER_ID = 1006;
SELECT …
改用JOIN寫有同樣效果
FROM CUSTOMER_T AS C
JOIN ORDER_T AS O ON C.CUSTOMER_ID = O.CUSTOMER_ID
JOIN ORDER_LINE_T AS L ON O.ORDER_ID = L.ORDER_ID
JOIN PRODUCT_T AS P ON L.PRODUCT_ID = P.PRODUCT_ID
WHERE ORDER_T.ORDER_ID = 1006;
Chapter 8
13
Figure 8-4 Results from a four-table join
From CUSTOMER_T table
From ORDER_T table
Chapter 8
From PRODUCT_T table
14
Self-Join Example
The same table is used
on both sides of the
join; distinguished
using table aliases
Self-joins are usually used on tables with unary relationships.
Chapter 8
15
Figure Example of a self-join
Chapter 8
16
Processing Multiple Tables
Using Subqueries

Subquery 因為查詢的結果還是表格，因此可對結果再查詢


Options:




placing an inner query (SELECT statement) inside
In a condition of the WHERE clause
As a “table” of the FROM clause
In the HAVING clause
Subqueries can be:


Noncorrelated–executed once for the entire outer query
Correlated–executed once for each row returned by the
outer query 每行資料都得執行一次子查詢
Chapter 8
17
Subquery Example

Show all customers who have placed an order
The IN operator will test to see if the
CUSTOMER_ID value of a row is
included in the list returned from the
subquery
SELECT CUSTOMER_NAME
FROM CUSTOMER_T
WHERE CUSTOMER_ID IN
(SELECT DISTINCT CUSTOMER_ID FROM ORDER_T);
Subquery is embedded in
parentheses. In this case it
returns a list that will be used
in the WHERE clause of the
outer query
Chapter 8
18
Join vs. Subquery

Some queries could be accomplished by either
a join or a subquery
Join version
Subquery version
Chapter 8
19
Figure Graphical depiction of two ways to answer a
query with different types of joins
Chapter 8
20
Figure Graphical depiction of two ways to answer a
query with different types of joins
Chapter 8
21
Correlated vs. Noncorrelated
Subqueries

Noncorrelated subqueries:



Do not depend on data from the outer query
Execute once for the entire outer query
Correlated subqueries:


Make use of data from the outer query
Execute once for each row of the outer query

Chapter 8
Can use with EXISTS operator 可搭配使用
22
Figure 8-6a Processing a noncorrelated subquery
No reference to data in outer query,
so subquery executes once only
Chapter 8
These are the only customers that
have IDs in the ORDER_T table
23
Correlated Subquery Example

Show all orders that include furniture finished in
natural ash
The EXISTS operator will return a
TRUE value if the subquery resulted
in a non-empty set, otherwise it
returns a FALSE
 A correlated subquery always refers to
an attribute from a table referenced in
the outer query
Chapter 8
The subquery is testing
for a value that comes
from the outer query
24
Figure 8-6b
Processing a
correlated
subquery
Subquery refers to
outer-query data,
so executes once
for each row of
outer query (需花
較多執行時間)
Chapter 8
25
Another Subquery Example

Show all products whose standard price is higher than
the average price
SELECT PRODUCT_DESCRIPTION, STANDARD_PRICE
FROM PRODUCT_T
WHERE STANDARD_PRICE >
(SELECT AVG(STANDARD_PRICE) AVGPRICE FROM PRODUCT_T)
Chapter 8
26
Union Queries

Combine the output (union of multiple queries)
together into a single result table
First query
Combine
Second query
Chapter 8
27
Tips for Developing Queries







Be familiar with the data model (entities and
relationships)
Understand the desired results
Know the attributes desired in result
Identify the entities that contain desired attributes
Review ERD
Construct a WHERE for each link 知道去哪查表
Fine tune with GROUP BY and HAING clauses if
needed
Chapter 8
28
Guidelines for Better Query Design








Write simple queries 越簡單越好
Break complex queries into multiple simple parts
把複雜查詢做拆解
If possible, avoid subquery and self-joins
Create temporary tables for groups of queries
Retrieve only the data you need i.e.不取多餘的
欄位或資料
Consider the total query processing time
Don’t have the DBMS sort without an index
Learn and practice 對複雜查詢多試不同的寫法
Chapter 8
29
Routines and Triggers

Routines



Program modules that execute on demand
Include Functions and Procedures
Ex. 預先寫好的常用SQL指令
Triggers

Routines that execute in response to a
database event (INSERT, UPDATE, or
DELETE)
Ex. 當INSERT至ORDER表格時，自動也
INSERT至ORDER_LOG表格
Chapter 8
30
Figure 8-10 Triggers contrasted with stored procedures
Procedures are called explicitly
Source: adapted from Mullins, 1995.
Chapter 8
Triggers are event-driven
31
Figure 8-11 Simplified trigger syntax, SQL:2008
Figure 8-12 Create routine syntax, SQL:2008
Chapter 8
32
Conditional Expressions Using Case Syntax
This is available with
newer versions of
SQL, previously not
part of the standard
Chapter 8
33
Embedded and Dynamic SQL

Embedded SQL


Including SQL statements in a program
將SQL指令放在C或Java程式內一起使用
Dynamic SQL

use program to generate SQL code on the fly
於程式內即時產生所需的SQL指令

Ex. 輸入客戶名稱檢查是否存在
SELECT count(*) FROM CUSTOMER WHERE NAME=$var_customer_name
Chapter 8
34
Ensuring Transaction Integrity

Transaction = A discrete unit of work that
must be completely processed or not
processed at all 確保動作完成不被中斷分割



May involve multiple updates
If any update fails, then all other updates must be
cancelled
SQL commands for transactions

BEGIN TRANSACTION/END TRANSACTION


COMMIT


Marks boundaries of a transaction
Makes all updates permanent
ROLLBACK

Chapter 8
Cancels updates since the last COMMIT
35
Figure 8-9 An SQL Transaction sequence (in pseudocode)
Chapter 8
36
Figure 11-9 Basic recovery techniques
a) Rollback
Chapter 11
37
Figure 11-9 Basic recovery techniques (cont.)
b) Rollforward
Chapter 11
38
Chapter 11
39
Why do we need transaction


When multiple users access the database…
Query the total balance

Transfer $100 from bank
A to B


SELECT sum(amount)
FROM account WHERE
id=‘001’
A tentative value will be
retrieved

Timeline
Chapter 8
UPDATE account SET
amount=amount-100
WHERE id=‘001’ and
bank=‘A’
UPDATE account SET
amount=amount+100
WHERE id=‘001’ and
bank=‘B’
40
Transaction ACID properties

4 properties that guarantee that database
transactions are processed reliably

Atomicity 不可分割性


Consistency 一致性


Constraints don't change from before transaction to after
transaction. i.e. database integrity remains.
Isolation 隔離性


Transaction cannot be subdivided; each transaction be "all or
nothing“
Database changes not revealed to users until after
transaction has completed; tentative data can't be accessed.
Durability 持續性

Chapter 8
Database changes are permanent; once a transaction has
been committed, it will remain so, even in the event of
power loss, crashes, or errors
41
CONTROL CONCURRENT ACCESS
Problem–in a multi-user environment,
simultaneous access to data can result in
interference and data loss (lost update
problem)
 Solution–Concurrency Control

 The
process of managing simultaneous
operations against a database so that data
integrity is maintained and the operations do
not interfere with each other in a multi-user
environment
 在網路、分散環境中特別重要且困難 →所以要用DBMS
Chapter 11
© 2013 Pearson Education
42
Figure 11-10 Lost update (no concurrency control in effect)
Simultaneous access causes updates to cancel each other.
A similar problem is the inconsistent read problem.
Chapter 11
© 2013 Pearson Education
43
CONCURRENCY CONTROL
TECHNIQUES

Serializability 循序存取


Finish one transaction before starting another
Locking Mechanisms 鎖定



The most common way of achieving
serialization
Data that is retrieved for the purpose of
updating is locked for the updater
No other user can perform update until
unlocked
Chapter 11
© 2013 Pearson Education
44
Figure 11-11: Updates with locking (concurrency control)
This prevents the lost update problem
Chapter 11
© 2013 Pearson Education
45
LOCKING MECHANISMS

Locking level: 鎖定的層級 (鎖越高可用性越低)






Database–used during database updates
Table–used for bulk updates
Block or page–very commonly used
Record–only requested row; fairly commonly used
Field–requires significant overhead; impractical
Types of locks:


Shared lock 共享–Read but no update permitted.
Used when just reading to prevent another user
from placing an exclusive lock on the record
Exclusive lock 互斥–No access permitted. Used
when preparing to update
Chapter 11
© 2013 Pearson Education
46
DEADLOCK

An impasse that results when two or more transactions have
locked common resources, and each waits for the other to
unlock their resources 互相擁有部分對方所需資源的僵局
Figure 11-12
The problem of deadlock
John and Marsha will wait
forever for each other to
release their locked
resources!
Chapter 11
© 2013 Pearson Education
47
MANAGING DEADLOCK

Deadlock prevention:


Lock all records required at the beginning of a
transaction
Two-phase locking protocol 兩階段鎖定法
Growing phase
 Shrinking phase



May be difficult to determine all needed resources in
advance
Deadlock Resolution:


Allow deadlocks to occur
Mechanisms for detecting and breaking them

Resource usage matrix 犧牲交易後重來
Chapter 11
© 2013 Pearson Education
48
VERSIONING 版本控制






Optimistic approach to concurrency control
Instead of locking 樂觀策略, 不用鎖定
Assumption is that simultaneous updates
will be infrequent 假設同時更新是較少發生的
Each transaction can attempt an update as it
wishes 所以先更新再說; 有衝突時再解決
The system will reject an update when it
senses a conflict
Use of rollback and commit for this 用
rollback解決衝突 (選擇犧牲者)
Chapter 11
© 2013 Pearson Education
49
Figure 11-14 The use of versioning
Better performance than locking
Chapter 11
© 2013 Pearson Education
50