How do I intentionally fragment a SQL Server Index?

Question

I would like to create bad indexes conditions intentionally on a SQL Server 2017 test database I have, just to better understand these maintenance scripts? SQL Server Index and Statistics Maintenance

Is there a fast/automatic way to compromise index integrity or increase index fragmentation? Do you know any useful resource I can look at to achieve this?

Depending on your definition of ugly you may want to mess up the fill factor as well which won't mess up fragmentation, but will have a degrading affect — S3S, Aug 19, 2019 at 19:15
Do you want for one index or all indexes in a database ? If you want for all indexes, then shrink your database - DBCC SHRINKDATABASE ([yourNONProdDB]) — Kin Shah, Aug 19, 2019 at 19:20

Evaldas Buinauskas · Accepted Answer · 2019-08-19 18:03:51Z

10

One quick way I can imagine is creating a table with UNIQUEIDENTIFIER as a primary key and inserting lots of random values. This could be achieved using this script:

CREATE TABLE dbo.Tests (Id UNIQUEIDENTIFIER PRIMARY KEY);
GO
INSERT INTO dbo.Tests (Id)
WITH x AS (SELECT n FROM (VALUES (0),(1),(2),(3),(4),(5),(6),(7),(8),(9)) v(n))
SELECT NEWID()
FROM x AS x1, x AS x2, x AS x3, x AS x4, x AS x5, x AS x6;

This will generate million rows.

Knowing that NEWID() does not guarantee any ordering, SQL Server will have to insert into random spots in table - that's going to fragment the primary key.

edited Aug 19, 2019 at 18:03

answered Aug 19, 2019 at 15:57

Evaldas Buinauskas

1,02011 silver badges22 bronze badges

1

That won't work or at least isn't guaranteed to: SQL Server can and probably will sort the rows (spooling to disk if needed due to the number of rows) before adding them to the index. To force fragmentation you need to perform many individual inserts - in SSMS: INSERT dbo.Tests (Id) SELECT NEWID(); GO 1000000; - this will obviously take more time. See pastebin.com/SVLtiRnP for an example I chucked together for another question. Using variable length rows and updating them at random might more efficiently produce fragmentation?
– David Spillett
Aug 20, 2019 at 10:40

Add a comment |

James Jenkins · Accepted Answer · 2019-08-19 16:09:50Z

I wanted to make several "Ugly" indexes, so I did the following. It worked well

-- Create databases to test index job, each database is about 800MB with 100,000 GUID primary keys, in each of two tables
-- Create 6 database to test index job for DatabasesInParallel Database design based on example https://dba.stackexchange.com/q/9821/21924

--Drop last test
USE [master]
exec asp_kill_user_connections [IndexTest_1]
exec asp_kill_user_connections [IndexTest_2]
exec asp_kill_user_connections [IndexTest_3]
exec asp_kill_user_connections [IndexTest_4]
exec asp_kill_user_connections [IndexTest_5]
exec asp_kill_user_connections [IndexTest_6]
GO

DROP DATABASE [IndexTest_1]
GO
DROP DATABASE [IndexTest_2]
GO
DROP DATABASE [IndexTest_3]
GO
DROP DATABASE [IndexTest_4]
GO
DROP DATABASE [IndexTest_5]
GO
DROP DATABASE [IndexTest_6]
GO

-- create [IndexTest_1]
USE [master];
GO

CREATE DATABASE [IndexTest_1];
GO

USE IndexTest_1

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END

-----------------------------------------------

-- create [IndexTest_2]
USE [master];
GO

CREATE DATABASE [IndexTest_2];
GO

USE IndexTest_2

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END

------------------------------------

-- create [IndexTest_3]
USE [master];
GO

CREATE DATABASE [IndexTest_3];
GO

USE IndexTest_3

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END

----------------------------------------
-- create [IndexTest_4]
USE [master];
GO

CREATE DATABASE [IndexTest_4];
GO

USE IndexTest_4

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END

------------------------------------------------
-- create [IndexTest_5]
USE [master];
GO

CREATE DATABASE [IndexTest_5];
GO

USE IndexTest_5

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END
--------------------------------------------

-- create [IndexTest_6]
USE [master];
GO

CREATE DATABASE [IndexTest_6];
GO

USE IndexTest_6

SET NOCOUNT ON

CREATE TABLE TestGuidA (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

CREATE TABLE TestGuidB (Id UNIQUEIDENTIFIER NOT NULL DEFAULT NEWID() PRIMARY KEY,
SomeDate DATETIME, batchNumber BIGINT, FILLER CHAR(100))

DECLARE @BatchCounter INT = 1
DECLARE @Numrows INT = 100000


WHILE (@BatchCounter <= 20)
BEGIN 
BEGIN TRAN

DECLARE @LocalCounter INT = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidA (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @LocalCounter = 0

    WHILE (@LocalCounter <= @NumRows)
    BEGIN
    INSERT TestGuidB (SomeDate,batchNumber) VALUES (GETDATE(),@BatchCounter)
    SET @LocalCounter +=1
    END

SET @BatchCounter +=1
COMMIT 
END
-------------------------------
use master
DBCC FREEPROCCACHE -- Clear plan cache for next text.

Shekar Kola · Accepted Answer · 2019-08-19 15:58:20Z

1

Usually, Index fragmentation happens when there is Update or Insert operation happens on the table.

If you want quickly produce the issue (Index fragmentation), create an Index in your test table with less fill factor and do heavy Update or Insert operation on that table. You can work with these scripts..

answered Aug 19, 2019 at 15:58

Shekar Kola

2,3972 gold badges9 silver badges23 bronze badges

Add a comment |

Tom V · Accepted Answer · 2019-08-20 10:01:30Z

1

You can also use CRYPT_GEN_RANDOM as I did in this answer: Filter Schema In Index Optimize Script.

You can insert data in a numeric column which has an index on it to fragment it like this:

-- Fill with random integers to create fragmentation
INSERT INTO [ProdTable] (c1, c2) VALUES  (CRYPT_GEN_RANDOM(8000), 'filler');
GO 12800

You could also update data or convert to a string instead of a number if that's what you need.

answered Aug 20, 2019 at 10:01

Tom V

15.6k7 gold badges62 silver badges86 bronze badges

Add a comment |

Stack Exchange Network

How do I intentionally fragment a SQL Server Index?

4 Answers 4

Your Answer

Not the answer you're looking for? Browse other questions tagged
sql-server
index
clustered-index
or ask your own question.

Linked

Hot Network Questions

How do I intentionally fragment a SQL Server Index?

4 Answers 4

Your Answer

Sign up or log in

Post as a guest

Not the answer you're looking for? Browse other questions tagged sql-serverindexclustered-index or ask your own question.

Linked

Related

Hot Network Questions

Not the answer you're looking for? Browse other questions tagged
sql-server
index
clustered-index
or ask your own question.