Index

NGUYEN Hong Phuong

Email: [email protected]
Site: https://users.soict.hust.edu.vn/phuongnh

1
Content
 What is Index?
 What is the index database used for?
 The structure of the index
 Types of indexes
 How to use index database effectively?

2
3
What is Index?
 A data structure is used to locate and
fast access data in tables or views.
 One way to increase database query
performance, by reducing the amount
of access to memory during query
execution
 SQL Server provides two types of
indexes
 Clustered
 Non-clustered
4
What is the index database used for?

 Query: SELECT * FROM student WHERE

last_name = 'May'
 If there is no index for the last_name
column, the system will scan all the rows
of the 'student' table to compare and
retrieve the row that satisfies
student

5
 An index points to the address of data
in a table, similar to a book's table of
contents, making queries fast
 Index can be created for one or more
columns in a table. Indexes are usually
created by default for primary keys,
foreign keys. In addition, it is also
possible to create additional indexes for
columns if needed.

6
The structure of the index
 Index includes:
 Search Key column: contains a copy of the
indexed column's values
 Data Reference column: contains the
pointer to the address of the record with
the corresponding index column value

7
Types of indexes
 B-tree
 Hash

8
B-tree
 Usually, if you don't specify the index type, the
default is to use B-Tree.
 Syntax:
 Create index
CREATE INDEX id_index ON table_name
(column_name [, column_name…]) USING BTREE;
ALTER TABLE table_name ADD INDEX id_index
(column_name [, column_name…])

 Delete the index

DROP INDEX index_name ON table_name

9
B-tree
 Features of B-Tree Index:
 Index data is organized and stored in the
form of tree, ie root, branch, leaf.
 The values of the organized nodes increase
from left to right.
 The B-tree index is used in comparison
expressions: =,>,> =, <, <=, BETWEEN,
and LIKE. Possible good for the ORDER
BY statement

10
B-tree
 When searching for data, it will not scan
the entire table. A search in B-Tree is a
process that starts from the root node and
searches for the branch and leaf, until
finding all data satisfying the query
condition.

11
Hash
 Hash index is based on Hash Function
algorithm. Corresponding to each block of
data, index will generate a bucket key
(hash value) to distinguish.
 Syntax:
 Create index
CREATE INDEX id_index
ON table_name(column_name [, column_name…]) USING HASH;

ALTER TABLE table_name

ADD INDEX id_index(column_name [, column_name…]) USING HASH;

12
Hash
 The features of Hash Index:
 Hash index should be used only in operator
'=' and '<>'. Do not use for operators to
find a range of values such as> or <.
 The ORDER BY operator cannot be
optimized using the Hash index because it
cannot find the next element in the Order.
 Hash is faster than B-Tree type.

13
Hash

14
Storage Engine
 Choosing the index of B-Tree or Hash
type, apart from the purpose of use, also
depends on whether or not the Storage
Engine supports the type of index.
 Storage Engine and index types are
supported
 InnoDB BTREE
 MyISAM BTREE
 MEMORY/HEAP HASH, BTREE
 NDB HASH, BTREE

15
How to use Index Database effectively?

 Should index the columns which are used in

WHERE, JOIN and ORDER BY
 Do not use index in the following cases: Không dùng
 Small tables, containing little data index nếu
 Tables are updated and data inserted regularly
 Columns that contain so many NULL values
 Columns are regularly updated
 Although index plays an important role in query
optimization and speeding up in searching in the
database, its downside is that it takes up more
memory to store. Therefore, indexing of columns
should be carefully considered
16
Practice with SQL Server
 BikeStores sample database

17
 Create a new 'production.parts' table
CREATE TABLE production.parts(
part_id INT NOT NULL,
part_name VARCHAR(100)
);

INSERT INTO
production.parts(part_id, part_name)
VALUES
(1,'Frame'),
(2,'Head Tube'),
(3,'Handlebar Grip'),
(4,'Shock Absorber'),
(5,'Fork');
18
 The 'parts' table doesn't have a
PK, so the records are stored in
an ordered structure called a SELECT
heap. part_id, part_name
FROM
 The statement finds records production.parts
with id 5 WHERE
part_id = 5;
 See execution plan estimates in
SQL Server Management Studio
 Select Display Estimated Execution
Plan (press Ctrl + L)

19
Two types of indexes in SQL Server

 Clustered index
 Non-clustered index

20
Clustered index
 Stores the records in an ordered
structure based on its key value
 Each table has only one clustered index
because the data records can only be
arranged in one order Mỗi bảng chỉ có 1
CLustered index
 A table that has a clustered index is
called a clustered table
 Data in clustered index are organized
under the form of B-tree
21
Clustered index

22
Clustered index
 The root node and the intermediate
node contain index pages for storing
the indexes of the records
 The leaf node contains the data pages
of the table.
 Pages within each level of the index are
linked in a double linked list structure

23
Clustered index
 When creating a table with the primary key PK, SQL Server
automatically creates a clustered index on the PK columns.
 The statement creates a 'part_prices' table with a PK
consisting of 2 columns:

CREATE TABLE production.part_prices(

part_id int,
valid_from date,
price decimal(18,4) not null,
PRIMARY KEY(part_id, valid_from)
);

24
Clustered index
 If you add the primary key to a table that already
has a clustered index, SQL Server forces the PK
to use the non-clustered index
ALTER TABLE
production.parts
ADD PRIMARY KEY(part_id);

25
 Clustered index
 Create clustered index
 In the case the table does not have a PK
CREATE CLUSTERED INDEX index_name
ON schema_name.table_name (column_list);
CREATE CLUSTERED INDEX ix_parts_id
ON production.parts (part_id);

 Query
SELECT part_id,
part_name
FROM production.parts
WHERE part_id = 5;
26
Non-clustered index
 Data structure that improves the speed of
retrieving data from tables
 Different from clustered index: Sorts and stores
data separately from records in the table.
 Is the data copy of selected columns from a
linked table.
 Use a B-tree structure to organize data
 A table can have one or more non-clustered
indexes. Each non-clustered index can consist of
one or more table columns.

27
Non-clustered index

28
Non-clustered index
 In addition to storing the index key
values, the leaf nodes also store
pointers to the records containing the
key values.
 These record pointers are also known
as row locators.

29
Non-clustered index
 Create non-clustered index
CREATE [NONCLUSTERED] INDEX index_name
ON table_name(column_list);
 The 'customers' table is a
clustered table because it
has the customer_id PK

30
Non-clustered index
 Search for customers whose address is at
'Atwater' SELECT customer_id, city
FROM sales.customers
WHERE city = 'Atwater';
 See execution plan estimation, the query
optimizer scans the clustered index for records,
since the 'customers' table doesn't have an index
for the 'city' column.
 Type the following command, and then see the
estimation again
CREATE INDEX ix_customers_city
ON sales.customers(city);

31
Non-clustered index
 Create non-clustered indexes for multiple
columns
 Find a customer with the last name 'Berg' and first name
'Monika'
SELECT customer_id, first_name, last_name
FROM sales.customers
WHERE last_name = 'Berg' AND first_name = 'Monika';

 See execution plan estimation, type the

following command, and run the above step
again
CREATE INDEX ix_customers_name
ON sales.customers(last_name, first_name);
32
Rename index in SQL Server
 The statement uses the sp_rename stored
procedure
EXEC sp_rename index_name, new_index_name, N'INDEX';

EXEC sp_rename @objname = N'index_name', @newname =

N'new_index_name', @objtype = N'INDEX';

 Ví dụ:
EXEC sp_rename
@objname = N'sales.customers.ix_customers_city',
@newname = N'ix_cust_city' ,
@objtype = N'INDEX';

EXEC sp_rename
N'sales.customers.ix_customers_city',
N'ix_cust_city' , 33
N'INDEX';
Rename index in SQL Server
 Or use SQL Server Management Studio,
right click,….

34
Unique index in SQL Server
 Unique indexes can be one or more
columns.
 If a column, the values in the column are
unique
 If more than one column, the combination of
values in these columns is unique
 Unique indexes can be clustered or non-
clustered indexes
 Syntax:
CREATE UNIQUE INDEX index_name
ON table_name(column_list);
35
Unique index
 For example, create a unique index for
an email column
 First of all, check to make sure there are
no duplicate email addresses
SELECT email, COUNT(email)
FROM sales.customers
GROUP BY email
HAVING COUNT(email) > 1;

CREATE UNIQUE INDEX ix_cust_email

ON sales.customers(email);

36
Unique index
 Try creating a table with 2 columns,
then create a unique index on both of
those columns
 Then, insert the data
 Is it okay if applying a unique index on
a column with multiple NULL values?
 Unique index vs. unique constraint

37
Disable indexing in SQL Server
 Before updating the table, disabling the index
speed up this process
ALTER INDEX index_name
ON table_name
DISABLE;
 Disable all indexes
ALTER INDEX ALL ON table_name
DISABLE;

38
Disable indexing in SQL Server
 If an index is disabled, the optimizer will not use
that index to plan the query execution
 Disabling indexing on a table, SQL Server retains
index definition in metadata and index statistics
in non-clustered indexes.
 Disable index on view, SQL Server will delete all
index data
 If a clustered index of a table is disabled, the
data of the table cannot be accessed using
SELECT, INSERT, UPDATE, and DELETE until the
clustered index is rebuilt / deleted.

39
Disable indexing in SQL Server

 Example:
ALTER INDEX ix_cust_city
ON sales.customers
DISABLE; ALTER INDEX ALL
ON sales.customers
SELECT DISABLE;
first_name,
last_name, SELECT * FROM sales.customers;
city
FROM
sales.customers
WHERE
city = 'San Jose';

40
Enable indexes in SQL Server
 After disabling the index for UPDATE, it
is necessary to re-enable the index
 The index needs to be rebuilt to reflect the
new data in the table
 Use one of the following two commands
 ALTER INDEX
 DBCC DBREINDEX

41
Enable indexes in SQL Server
 ALTER INDEX and CREATE INDEX
ALTER INDEX index_name
ON table_name
REBUILD;

CREATE INDEX index_name

ON table_name(column_list)
WITH(DROP_EXISTING=ON)

ALTER INDEX ALL ON table_name

REBUILD;

42
Enable indexes in SQL Server
 DBCC DBREINDEX

DBCC DBREINDEX (table_name, index_name);

ALTER INDEX ALL ON sales.customers

REBUILD;

43
Delete indexes in SQL Server
 DROP INDEX
DROP INDEX [IF EXISTS] index_name
ON table_name;
 The DROP INDEX statement cannot delete
indexes created by PK or a unique constraint
 To remove indexes associated with these constraints,
use the ALTER TABLE DROP CONSTRAINT command
 Remove multiple indexes from one/multiple
tables, use the following command:
DROP INDEX [IF EXISTS]
index_name1 ON table_name1,
index_name2 ON table_name2,
...; 44
Filtered index in SQL Server
 Sometimes, it is inefficient to index all the
records by a certain column, as it is only partially
queried for a few of records of the whole table.
 A filtered index is a non-clustered index with an
expression that specifies which records should be
added to the index.
 Syntax:
CREATE INDEX index_name
ON table_name(column_list)
WHERE predicate;

45
Filtered index in SQL Server
 For example, using the 'customers' table, the
phone column has so many NULL values
SELECT
SUM(CASE CREATE INDEX ix_cust_phone
WHEN phone IS NULL ON sales.customers(phone)
THEN 1 WHERE phone IS NOT NULL;
ELSE 0
END) AS [Has Phone],
SUM(CASE SELECT
WHEN phone IS NULL first_name,
THEN 0 last_name,
ELSE 1 phone
END) AS [No Phone] FROM sales.customers
FROM sales.customers; WHERE phone = '(281) 363-3309';

46
Filtered index in SQL Server
 INCLUDE
CREATE INDEX ix_cust_phone
ON sales.customers(phone)
INCLUDE (first_name, last_name)
WHERE phone IS NOT NULL;

47
48