SQL & ADVANCED SQL

Marcin Blaszczyk (CERN IT-DB)

marcin.blaszczyk@cern.ch
 AGENDA
 Goal of this tutorial:
 Present the overview of basic SQL capabilities
 Explain several selected advanced SQL features

 Outline
 Introduction
 SQL basics
 Joins & Complex queries
 Analytical functions & Set operators
 Other DB objects (Sequences, Synonyms, DBlinks, Views & Mviews)
 Indexes & IOTs
 Partitioning
 Undo & Flashback technologies

Oracle Tutorials 5th of May 2012

 SQL LANGUAGE
 Objective: be able to perform the basic operation of the
RDBMS data model
 create, modify the layout of a table
 remove a table from the user schema
 insert data into the table
 retrieve and manipulate data from one or more tables
 update/ delete data in a table
 +
 Some more advanced modifications

Oracle Tutorials 5th of May 2012

 SQL LANGUAGE (2)
 Structured Query Language
 Programing language
 Designed to mange data in relational databases

 DDL Data Definition Language

 Creating, replacing, altering, and dropping objects
 Example: DROP TABLE [TABLE];
 DML Data Modification Language
 Inserting, updating, and deleting rows in a table
 Example: DELETE FROM [TABLE];
 DCL Data Control Language
 Controlling access to the database and its objects
 Example: GRANT SELECT ON [TABLE] TO [USER];

Oracle Tutorials 5th of May 2012

 SQL LANGUAGE(3)
STATEMENT DESCRIPTION
SELECT Data Retrieval

INSERT
UPDATE Data Manipulation Language (DML)
DELETE

CREATE
ALTER
DROP Data Definition Language (DDL)
RENAME
TRUNCATE

GRANT
Data Control Language (DCL)
REVOKE

COMMIT
Transaction Control
ROLLBACK

Oracle Tutorials 5th of May 2012

 TRANSACTION & UNDO
 A transaction is a sequence of SQL Statements that Oracle treats
as a single unit of work
 A transaction must be commited or rolled back:
COMMIT; - makes permanent the database changes you made during the
transaction.
ROLLBACK; - ends the current transaction and undoes any changes
made since the transaction began.
 Check COMMIT settings in your Client Tool (eg
AUTOCOMMIT, EXITCOMMIT in SQL*Plus)
 UNDO tablespace:
 circular buffer
 records all actions of transactions
 used when rolling back a transaction

Oracle Tutorials 5th of May 2012

 SQL LANGUAGE(3)
STATEMENT DESCRIPTION
SELECT Data Retrieval

INSERT
UPDATE Data Manipulation Language (DML)
DELETE

CREATE
ALTER
DROP Data Definition Language (DDL)
RENAME
TRUNCATE

GRANT
Data Control Language (DCL)
REVOKE

COMMIT
Transaction Control
ROLLBACK

Oracle Tutorials 5th of May 2012

 DATABASE SCHEMA (USER)
 Collection of logical structures of data
 called schema objects
 tables, views, indexes, synonyms, sequences, packages, triggers,
links, …

 Owned by a database user

 same name of the user

 Schema objects can be created and manipulated with SQL

SELECT * FROM USER_OBJECTS | USER_TABLES (…)

SELECT user DROM dual;

SHOW USER; (in SQL*Plus)

Oracle Tutorials 5th of May 2012

 CREATE A TABLE
 Define the table layout:
 table identifier
 column identifiers and data types
 column constraints, CREATE TABLE employees (
employee_id NUMBER(6) NOT NULL,
 default values first_name VARCHAR2(20),
last_name VARCHAR2(25),
 integrity constraints hire_date DATE DEFAULT SYSDATE,
department_id NUMBER(4),
 relational constraints salary NUMBER(8,2) CHECK (salary > 0));
SQL> describe employees
Name Null? Type
----------------- -------- ------------
EMPLOYEE_ID NOT NULL NUMBER(6)
FIRST_NAME VARCHAR2(20)
LAST_NAME VARCHAR2(25)
HIRE_DATE DATE
DEPARTMENT_ID NUMBER(4)
SALARY NUMBER(8,2)

Oracle Tutorials 5th of May 2012

 DATATYPES
 Each value has a datatype
 defines the domain of values that each column can contain
 when you create a table, you must specify a datatype for each
of its columns

 ANSI defines a common set ANSI data type Oracle

integer NUMBER(38)
 Oracle has its set of built-in types
smallint NUMBER(38)
 User-defined types
numeric(p,s) NUMBER(p,s)

varchar(n) VARCHAR2(n)

char(n) CHAR(n)

float NUMBER

real NUMBER

Oracle Tutorials 5th of May 2012

 SELECT STATEMENT
SELECT [ALL | DISTINCT] column1[,column2]
FROM table1[,table2]
[WHERE "conditions"]
[GROUP BY "column-list"]
[HAVING "conditions]
[ORDER BY "column-list" [ASC | DESC] ]
SELECT d.department_name,
sum(e.salary)as DEPT_AL DEPARTMENT_NAME DEPT_SAL
FROM departments d, employees e -------------------- ----------
WHERE d.department_id = e.department_id Accounting 20300
GROUP BY d.department_name Executive 58000
HAVING SUM(e.salary) > 10000 Finance 51600
ORDER BY department_name; IT 28800
Marketing 19000
Purchasing 24900
Sales 304500
Shipping 156400

Oracle Tutorials 5th of May 2012

 INSERT, UPDATE, DELETE (DML)
 Insert some data
INSERT INTO table1 values(value-list) ;
INSERT INTO table1(column-list) values(value-list);
INSERT INTO table1(column-list)
SELECT values(value-list);
COMMIT;
 Update
UPDATE table1 SET column = value;
COMMIT;
 Delete
DELETE FROM table1;
COMMIT;

Oracle Tutorials 5th of May 2012

 ALTER TABLE (DDL)
 Modify the name:
ALTER TABLE employees RENAME TO newemployees;

 Modify the layout:

ALTER TABLE employees ADD (salary NUMBER(7));
ALTER TABLE employees RENAME COLUMN id TO emp_id;
ALTER TABLE employees DROP(hiredate);

 But also:
 Add/modify/drop constraints
 Enable/Disable constraints
 Modify more advanced properties…

Oracle Tutorials 5th of May 2012

 CONSTRAINTS (DDL)
 NOT NULL / CHECK
ALTER TABLE employees MODIFY last_name NOT NULL;
ALTER TABLE employees MODIFY salary CHECK (salary > 1000);
 PRIMARY KEY
ALTER TABLE employees ADD PRIMARY KEY(emp_id);
 FOREIGN KEY
ALTER TABLE employees ADD FOREIGN KEY(dept_id) REFERENCES
departments(department_id);
 Constraints errors:
 ORA-02290: check constraint (owner.constraintname) violated – DURING
INSERT
 ORA-02291: integrity constraint (owner.constraintname) violated - parent key
not found – DURING INSERT
 ORA-02292:violated integrity constraint (owner.constraintname)- child record
found – DURING DELETE

Oracle Tutorials 5th of May 2012

 NULL VALUE
 special value that means
 unavailable
 unassigned
 unknown
 inapplicable
 not equivalent to
 zero
 blank space

SELECT * FROM [TABLE] where id = 0;

SELECT * FROM [TABLE] where id IS NULL;

 Often used as default

Oracle Tutorials 5th of May 2012

 DUAL TABLE
 special one-row table present by default in all Oracle database
installations SQL> describe dual;
 Accessible to all users Name Null? Type
 Examples of use: ----------------- -------- ------------
DUMMY VARCHAR2(1)

SELECT SYSDATE FROM DUAL;

SELECT USER FROM DUAL;
-- equal to SHOW USER in SQL*Plus

 Create really big table in one command - use dual;

CREATE TABLE BIG_TABLE

AS SELECT trunc(dbms_random.value(0,20)) RANDOM_INT
FROM DUAL
CONNECT BY LEVEL <= 100000;

Oracle Tutorials 5th of May 2012

 DELETE ALL ROWS FROM A TABLE
?

 What is the difference between:

DELETE FROM employees;
vs
TRUNCATE TABLE employees;

 DML vs DDL commands?

 Is COMMIT essential? In which case?
 Generate UNDO segments?
 Which is more efficient?

Oracle Tutorials 5th of May 2012

 TYPES OF JOINS

Values in the two corresponding columns of

EQUIJOIN
the different tables must be equal

NON-EQUIJOIN The relationship between the columns of the

different tables must be other than equal

OUTERJOIN It returns also the rows that do not satisfy the

(LEFT, RIGHT, FULL) join condition

SELFJOIN Joining data in a table to itself

Oracle Tutorials 5th of May 2012

 EQUIJOIN
SQL> SELECT e.emp_name, e.emp_deptno, d.dept_name
FROM emp e, dept d
WHERE e.emp_deptno = d.deptno
ORDER BY emp_name;

EMP_NAME EMP_DEPTNO
KING 10

BLAKE 30 DEPT_NO DEPT_NAME

10 ACCOUNTING
CLARK 10
30 SALES

20 OPERATIONS

EMP_NAME EMP_DEPTNO DEPT_NAME

KING 10 ACCOUNTING

BLAKE 30 SALES

CLARK 10 ACCOUNTING

Oracle Tutorials 5th of May 2012

 OUTERJOIN
SQL> SELECT e.emp_name, e.emp_deptno, d.dept_name
FROM emp e, dept d
WHERE e.emp_deptno = d.deptno(+)
ORDER BY emp_name;

EMP_NAME EMP_DEPTNO
DEPT_NO DEPT_NAME
KING 10
10 ACCOUNTING
BLAKE NULL
30 SALES
CLARK 10
20 OPERATIONS
MARTIN 20
EMP_NAME EMP_DEPTNO DEPT_NAME
TURNER 10
KING 10 ACCOUNTING
JONES NULL
BLAKE NULL NULL

CLARK 10 ACCOUNTING

MARTIN 20 OPERATIONS
TURNER 10 ACCOUNTING
JONES NULL NULL

Oracle Tutorials 5th of May 2012

 JOINS SYNTAX ANSI VS ORACLE
 Equijoins:
 ANSI syntax
SELECT e.name, d.name FROM employees e

INNER JOIN departments d ON e.dept_id=d.dept_id;

 Oracle
SELECT e.name, d.name FROM employees e, departments d

WHERE e.dept_id=d.dept_id;

 Outerjoins
 ANSI syntax (LEFT, RIGHT, FULL)
SELECT e.name, d.name FROM employees e

RIGHT OUTER JOIN departments d ON e.dept_id=d.dept_id;

 Oracle
SELECT e.name, d.name FROM employees e, departments d

WHERE e.dept_id(+)=d.dept_id;

Oracle Tutorials 5th of May 2012

 ADVANCED SQL QUERIES
Types Question

Who works in the same department as

SUBQUERIES
Clark?

Who are the employees that receive more

Correlated SUBQUERIES than the average salary of their
department?

What are the employees salary and the

Inline Views
minimum salary in their department?

What are the 5 most well paid

Top-N QUERIES
employees?

What is the hierarchy of management in

Hierarchical QUERIES
my enterprise?

Oracle Tutorials 5th of May 2012

 SUBQUERIES (1/5)
 A subquery is a query within a query and it is used to answer
multiple-part questions.

 Oracle fully supports them in the sense that:

 You can create subqueries within your SQL statements
 A subquery can reside in the WHERE clause, the FROM
clause or the SELECT clause.

Subquery Inline view Nested subquery

SELECT ... FROM ... WHERE ...

Oracle Tutorials 5th of May 2012

 SUBQUERIES (2/5)

A) Single-row (and single-column)

Types B) Multiple-row (and single-column)
C) Multiple-column

 who works in the same department as Clark?

SELECT … WHERE dep = (SELECT dep FROM … WHERE name = ‘CLARK’);

 who works in the same department as Clark OR Blake?

SELECT … WHERE dep IN (SELECT dep
FROM …
WHERE name =‘CLARK’ or name = ‘BLAKE’);
 who works in the same department(s) AND under the same boss as Clark?
SELECT … WHERE (dep, mgr) = (SELECT dep, mgr
FROM …
WHERE name = ‘CLARK’)

Oracle Tutorials 5th of May 2012

 CORRELATED SUBQUERIES
 A correlated subquery is a subquery that is evaluated FOR
EACH ROW produced by the parent query.

 Which employees receive more than the average salary of

their department? EMP_ID DEPT_ID LAST_NAME SALARY
------ ------- ---------- ------
201 20 Hartstein 13000
SELECT e.emp_id, e.dept_id, 114 30 Raphaely 11000
e.last_name, e.salary 123 50 Vollman 6500
122 50 Kaufling 7900
FROM employees e 120 50 Weiss 8000
WHERE e.salary > (SELECT avg(i.salary) 121 50 Fripp 8200
103 60 Hunold 9000
FROM employees i 147 80 Errazuriz 12000
WHERE e.dept_id = i.dept_id) 146 80 Partners 13500
145 80 Russell 14000
100 90 King 24000
108 100 Greenberg 12000

 In this case, the correlated subquery specifically computes,

for each employee, the average salary for the employee’s
department

Oracle Tutorials 5th of May 2012

 INLINE VIEWS
 An In-line view is a subquery in the FROM clause of a SQL
statement just as if it was a table. It acts as a data source!

 What are the employees salary and the MINIMAL salary in their
department?
SELECT e.emp_id a.dept_id, e.last_name, EMP_ID DEPT_ID LAST_NAME SALARY MIN_SAL
------ ------- ----------- ------ -------
e.salary, a.min_sal, 200 10 Whalen 4400 4400
FROM employees e, 201 20 Hartstein 13000 6000
202 20 Fay 6000 6000
(SELECT MIN(salary)min_sal, dept_id 114 30 Raphaely 11000 2500
FROM employees 115 30 Khoo 3100 2500
116 30 Baida 2900 2500
GROUP BY dept_id) a 117 30 Tobias 2800 2500
WHERE e.dept_id = a.dept_id 118 30 Himuro 2600 2500
119 30 Colmenares 2500 2500
ORDER BY e.dept_id, e.salary DESC; 203 40 Mavris 6500 6500
121 50 Fripp 8200 2100
120 50 Weiss 8000 2100
122 50 Kaufling 7900 2100
123 50 Vollman 6500 2100
124 50 Mourgos 5800 2100
184 50 Sarchand 4200 2100
185 50 Bull 4100 2100
192 50 Bell 4000 2100

Oracle Tutorials 5th of May 2012

 TOP-N QUERIES
 We need to use “in-line view” together with the ROWNUM
pseudocolumn
 What are the top 5 most well paid employees?
SELECT * FROM EMP_ID LAST_NAME SALARY
(SELECT emp_id, last_name, salary ------ ---------- ------
100 King 24000
FROM employees
101 Kochhar 17000
ORDER BY salary desc) 102 De Haan 17000
WHERE rownum < 6 145 Russell 14000
146 Partners 13500

 What are the next 5 most well paid employees?

SELECT emp_id, last_name, salary FROM ( EMP_ID LAST_NAME SALARY
SELECT emp_id,last_name, salary, ------ ---------- ------
rownum as rnum 108 Greenberg 12000
109 Faviet 9000
FROM employees 106 Pataballa 4800
ORDER BY salary desc) 105 Austin 4800
WHERE rnum between 6 and 10; 107 Lorentz 4200

Oracle Tutorials 5th of May 2012

 HIERARCHICAL QUERIES
 If a table contains hierarchical data, then you can select rows in a
hierarchical order using the hierarchical query clause
 Syntax:
SELECT … FROM … WHERE …
START WITH <condition>
Specifies the starting point of the hierarchy (tree)
CONNECT BY PRIOR child_row = parent_row (TOP-DOWN)
parent_row = child_row (BOTTOM-UP)
relationship between parent row and child rows of the hierarchy

 Pseudo-column LEVEL is the hierarchy level

King
SELECT empid, last_name, mgrid, LEVEL
FROM employees
WHERE LEVEL <= 3 Kochnar
START WITH employee_id = 100
CONNECT BY PRIOR EMPID LAST_NAME MGRID LEVEL
employee_id = manager_id; ----- --------- ----- -----
100 King 1 Whalen Mavris Baer
101 Kochhar 100 2
200 Whalen 101 3
203 Mavris 101 3
204 Baer 101 3

Oracle Tutorials 5th of May 2012

 HIERARCHICAL QUERIES
 If a table contains hierarchical data, then you can select rows in a
hierarchical order using the hierarchical query clause
 Syntax:
SELECT … FROM … WHERE …
START WITH <condition>
Specifies the starting point of the hierarchy (tree)
CONNECT BY PRIOR child_row = parent_row (TOP-DOWN)
parent_row = child_row (BOTTOM-UP)
relationship between parent row and child rows of the hierarchy

 Pseudo-column LEVEL is the hierarchy level

King
SELECT empid, last_name, mgrid, LEVEL
FROM employees
START WITH employee_id = 204 Kochnar
CONNECT BY PRIOR
manager_id = employee_id; EMPID LAST_NAM MGR_ID LEVEL
----- -------- ------ ------
204 Baer 101 1 Baer
101 Kochhar 100 2
100 King 3

Oracle Tutorials 5th of May 2012

 ANALYTICAL FUNCTIONS
 General syntax of analytical function: TABLE

SELECT analytical-function(col-expr)
EWWEQEWQEQ
WINDOW
OVER (window-spec) [AS col-alias]
FROM [TABLE]; CURRENT ROW

 Window specification syntax ASDASSDA

[PARTITION BY [expr list]]

ORDER BY [sort spec] [range spec]
 Example for range specification (for more check oracle docs)
ROWS UNBOUNDED PRECEDING AND CURRENT ROW (default)
ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING
RANGE BETWEEN 2 PRECEDING AND 2 FOLLOWING

Oracle Tutorials 5th of May 2012

 ORDERRED ANALYTICAL WINDOW
 Analytical functions applied to all window rows
 Remember about ordering inside the window
SQL> select employee_id, last_name, manager_id, salary
sum(salary) over (order by employee_id, last_name, salary)
as cumulative from employees;

EMPLOYEE_ID LAST_NAME MANAGER_ID SALARY CUMULATIVE

----------- --------- ---------- ------ ----------
100 King 24000 24000
101 Kochhar 100 17000 41000
102 De Haan 100 17000 58000 = 24000+17000+17000
103 Hunold 102 9000 67000
104 Ernst 103 6000 73000
105 Austin 103 4800 77800
106 Pataballa 103 4800 82600
107 Lorentz 103 4200 86800
108 Greenberg 101 12000 98800
109 Faviet 108 9000 107800
110 Chen 108 8200 116000

Oracle Tutorials 5th of May 2012

 RANGE SPECIFICATION (1/2)

 RANGE BETWEEN 2 PRECEDING AND 2 FOLLOWING

SQL> select manager_id, last_name, salary, sum(salary) over (order by
last_name, salary rows between 2 preceding and 1 following) as
cumulative from employees;

MANAGER_ID LAST_NAME SALARY CUMULATIVE

---------- --------- ------ ----------
103 Austin 4800 10800
103 Ernst 6000 22800
101 Greenberg 12000 31800
102 Hunold 9000 51000 = 6000 + 12000 + 9000 + 24000
King 24000 62000
100 Kochhar 17000 54200
103 Lorentz 4200 45200

Oracle Tutorials 5th of May 2012

 RANGE SPECIFICATION (2/2)

 ROWS BETWEEN CURRENT ROW AND UNBOUNDED FOLLOWING

SQL> select manager_id, last_name, salary, sum(salary) over (order by
last_name, salary rows between current row and unbounded following)
as cumulative from emp_part;

MANAGER_ID LAST_NAME SALARY CUMULATIVE

---------- --------- ------ ----------
103 Austin 4800 77000
103 Ernst 6000 72200
101 Greenberg 12000 66200
102 Hunold 9000 54200 = 9000 + 24000 + 17000 + 4200
King 24000 45200
100 Kochhar 17000 21200
103 Lorentz 4200 4200

Oracle Tutorials 5th of May 2012

 PARTITIONED ANALYTICAL WINDOW
 Analytical functions start again for each partition
SQL> break on manager_id
SQL> SELECT manager_id, last_name, employee_id, salary,
sum(salary) over (PARTITION BY manager_id order by employee_id)
as cumulative
FROM employees order by manager_id, employee_id, last_name;

MANAGER_ID LAST_NAME EMPLOYEE_ID SALARY CUMULATIVE

---------- --------------- ----------- ------ ----------
100 Kochhar 101 17000 17000
De Haan 102 17000 34000
Raphaely 114 11000 45000
Weiss 120 8000 53000
101 Greenberg 108 12000 12000
Whalen 200 4400 16400
Mavris 203 6500 22900
Baer 204 10000 32900
102 Hunold 103 9000 9000
103 Ernst 104 6000 6000
Austin 105 4800 10800
Pataballa 106 4800 15600

Oracle Tutorials 5th of May 2012

 ANALYTIC FUNCTIONS
 For analytic functions, you can use all of the regular group
functions
 SUM
 MAX
 MIN
 AVG
 COUNT
 Plus list of additional analytical functions that can be used only for
window queries.
 LAG
 LEAD
 FIRST
 LAST
 FIRST VALUE
 LAST VALUE
 ROW_NUMBER
 DENSE_RANK

Oracle Tutorials 5th of May 2012

 ANALYTICAL FUNCTION EXAMPLE
 LAG function example SQL> select day, EURCHF,
lag(EURCHF,1) over (order by day)
SQL> select * from currency order by 1; as prev_eurchf from currency;

DAY EURCHF DAY EURCHF PREV_EURCHF

-------------------- ------ -------------------- ------ --------------
01-JUN-2012 00:00:00 1.240 01-JUN-2012 00:00:00 1.240
02-JUN-2012 00:00:00 1.223 02-JUN-2012 00:00:00 1.223 1.240
03-JUN-2012 00:00:00 1.228 03-JUN-2012 00:00:00 1.228 1.223
04-JUN-2012 00:00:00 1.217 04-JUN-2012 00:00:00 1.217 1.228
05-JUN-2012 00:00:00 1.255 05-JUN-2012 00:00:00 1.255 1.217
06-JUN-2012 00:00:00 1.289 06-JUN-2012 00:00:00 1.289 1.255
07-JUN-2012 00:00:00 1.291 07-JUN-2012 00:00:00 1.291 1.289
08-JUN-2012 00:00:00 1.247 08-JUN-2012 00:00:00 1.247 1.291
09-JUN-2012 00:00:00 1.217 09-JUN-2012 00:00:00 1.217 1.247
10-JUN-2012 00:00:00 1.265 10-JUN-2012 00:00:00 1.265 1.217

SQL> select day, EURCHF, ((EURCHF - prev_eurchf) / prev_eurchf )*100 as pct_change from (
select day, EURCHF, LAG(EURCHF,1) over (order by day) as prev_eurchf from currency);

DAY EURCHF PCT_CHANGE

6
-------------------- ------ ----------
01-JUN-2012 00:00:00 1.240 4 3.94
02-JUN-2012 00:00:00 1.223 -1.37 3.12 2.71
03-JUN-2012 00:00:00 1.228 0.41 2
04-JUN-2012 00:00:00 1.217 -0.90 0.41
0 0.16
05-JUN-2012 00:00:00 1.255 3.12 -0.9
06-JUN-2012 00:00:00 1.289 2.71 -1.37
-2 -2.41
07-JUN-2012 00:00:00 1.291 0.16
-3.41
08-JUN-2012 00:00:00 1.247 -3.41 -4
09-JUN-2012 00:00:00 1.217 -2.41
10-JUN-2012 00:00:00 1.265 3.94

Oracle Tutorials 5th of May 2012

 SET OPERATORS
 Combine multiple queries

 Union without duplicates

SELECT name, email FROM employees
UNION
SELECT name, email FROM visitors;

 Union with the duplicates

SELECT cit_id FROM employees
UNION ALL
SELECT cit_id FROM visitors;

 Intersect
SELECT name FROM employees
INTERSECT
SELECT name FROM visitors;

 Minus
SELECT name FROM employees
MINUS
SELECT name FROM visitors;

Oracle Tutorials 5th of May 2012