When I usually write a blog post that is often due to questions I receive in my mailbox or real problems my clients face and I have to resolve. A few days ago I got an email from a friend of mine. He was telling me that his website was down and was not functioning. He said the error he was seeing in was

“The transaction log for database ‘mydb’ is full. To find out why space in the log cannot be reused, see the log_reuse_wait_desc column in sys.databases”

It is obvious that there is a problem with the transaction log and that is why his site is not functioning.I will also provide hands-on examples when needed.

Before we proceed I would like to explain the various recovery models.We have 3 recovery models available in SQL Server

• Simple
• Full
• Bulk-Logged

In the Simple Recovery Model every time I have a checkpoint operation or the .ldf file is full, the .ldf file is automatically truncated.In this model the transaction log is not backed up.You can not have point in time data recovery with this model.Data is recoverable only to the most recent backup.

The Simple Recovery Model  is only suitable for small databases and databases with few transactions.We could use this recovery model if we want to have a small transaction log and for databases that are read-only or in development stage.

The Full Recovery Model is the default recovery model.In this model the transaction log grows rapidly since all database transactions are logged in at a very detailed level.If I have a catastrophe in my database I can rollback to the time where the catastrophe occurred.This recovery model is suitable for databases with thousands of transactions per day and where no loss of data is tolerated.

The Bulk Recovery Model works almost identically with the Full Recovery Model.It logs everything in extents=8pages and all bulk tasks (bcp,bulk insert) are logged minimally.We can go back to the time of failure but it will take more time to get the database back on-line.This model would be very suitable for databases with large bulk tasks.The disadvantage of this model is that you cannot go back at a specific point in time.

Having looked into the recovery models,let’s have a look on the various SQL Server backup types.

Let’s see what happens when we take a full database backup.The first thing to say is that you need always a full database backup.This is the base for all subsequent backups. You cannot have a differential backup,transaction log backup if you do not have at least one full backup.With a full database backup you backup all your data.You can have a full backup with all the recovery models that you choose to have for your database.Every full backup I perform in a database contains in it all the data previously fully backed up.
The backup operation is an online operation and the first thing it happens is to log that a backup was started. Then we write the data in the backup file.Then we do log any changes that occur as we backup.Finally the backing up of the data is finished and we log that fact.If you’re performing only full backups, you might lose some data in the event of a system crash—specifically, any changes made since the last full backup.

A differential backup is cumulative backup. It includes all data and structures that have changed since the last full backup, regardless of when that last full backup was made, or how many previous differential backups have been run.You can have a differential backup with all the recovery models.

The Transaction Log backup only works for full or bulk-logged recovery models.Each T-Log backup contains all the transactions that are in my database since the last T-Log backup.Always perform a full back up before you take a T-log backup.The time taken for a T-log backup is little and there is no performance penalty.An important thing to remember is that for every t-log backup I need the previous backup to have a restore.

Let me come back to the original problem which is that my friend’s website run out of transaction log space.
This happens when the transaction log fills up to the point where it has to grow but either autogrow is not turned on or the volume on which the transaction log is placed has no more space for the file to grow.

That is profound. But what has caused the transaction log to fill up in the first place? It could be a number of different things.

The two most common ones I’ve seen are:

1. The database is in full recovery model and subsequent full database backups were taken but no log backups have been taken.

This commonly happens when an application goes into production and someone decides to take a backup, without realizing the consequences. Log truncation is not automatic in such a case – a log backup is required to allow the log to truncate.

2. The database is in any recovery mode but there’s a very long-running, uncommitted transaction that prevents log truncation.

In this case I was called to examine, the problem was that the database was in full recovery mode, subsequent full database backups were taken but no log backups have been taken.

Some people believe that a full database backup in the full or bulk logged will clear/truncate the transaction log.

Well, these people are wrong. Simple as that. They are wrong.This is not true. Only a transaction log backup will clear the log under these recovery models.

On the other hand, in the SIMPLE recovery model, a checkpoint clears the transaction log. I see people that have critical data in their databases and need to restore it (in case of a catastrophe) to the point of failure, to use Simple recovery model in their databases because the transaction log is getting too large. So in order to keep it small, they switch recovery models. This is so wrong!!!! In case of a failure these people will lose data they cannot afford to lose.

Let me go on with the actual example

I will be using SQL Server 2012 Εnterprise edition in this post but feel free to use SQL Server 2008/R2 versions and any other edition you might have installed (Standard, Express e.t.c).

1) Launch SSMS and connect to the local (or any other instance) of SQL Server

2) I will be looking at the column log_reuse_wait_desc of the sys.databases

3) In a new query window in SSMS , type (or copy paste) the following t-sql script and then execute it (substitute database football with any other you have attached in the instance of SQL Server.)

SELECT  log_reuse_wait_desc
FROM    sys.databases
WHERE   name = 'football'

In my case I receive “Nothing” and that is why you will receive yourself if you execute the T-SQL script above but in my friend’s case when I connected to the SQL hosting the database I go the following:

Log Backup

That is all I needed to know. SQL Server tells me what the problem is.The database was in Full recovery model, only full backups were taken and no log backups have been taken.

4) Now I will show you why a full database backup does not clear the transaction log

Type (copy and paste) the following T-SQL script but do not execute it yet.

USE MASTER;
GO

CREATE DATABASE [Football] ON  PRIMARY
( NAME = N'Football_Data', FILENAME = N'D:\DATA\Football_Data.mdf' , SIZE = 80MB , MAXSIZE = UNLIMITED, FILEGROWTH = 20MB )
LOG ON
( NAME = N'Football_Log', FILENAME = N'D:\DATA\Football_log.ldf' , SIZE = 20MB , MAXSIZE = UNLIMITED, FILEGROWTH = 5MB)
GO

USE FOOTBALL;
GO

CREATE TABLE Footballer
(
FootballerID int identity(1,1),
Footballer_No varchar(20),
Footballer_lastname nvarchar(50),
Footballer_firstname nvarchar(50),
Footballer_middlename nvarchar(50)
)
GO

CREATE UNIQUE CLUSTERED INDEX cx_Foo_IX ON Footballer (FootballerID)
GO

declare @CTR INT
declare @ctrstr varchar(7)
select @ctr=0
WHILE @CTR<4000
BEGIN
select @ctr=@ctr+1
select @ctrstr=Str(@ctr)
insert footballer(Footballer_No ,Footballer_lastname,Footballer_firstname, Footballer_middlename) values('F_No'+@ctrstr,'Footlname'+@ctrstr,'Footfname'+@ctrstr,'Footmname'+@ctrstr)
END

BACKUP DATABASE [Football] TO  DISK = N'D:\DATA\backup\football_full.bak' WITH INIT,STATS;
GO

BACKUP LOG [Football] TO  DISK = N'D:\DATA\backup\football_log.bak' WITH INIT,STATS;
GO

ALTER INDEX cx_Foo_IX ON Footballer REBUILD;
GO

BACKUP LOG [Football] TO  DISK = N'D:\DATA\backup\football1_log.bak' WITH INIT,STATS;
GO

ALTER INDEX cx_Foo_IX ON Footballer REBUILD;
GO

BACKUP DATABASE [Football] TO  DISK = N'D:\DATA\backup\football1_full.bak' WITH INIT,STATS;
GO

BACKUP LOG [Football] TO  DISK = N'D:\DATA\backup\football2_log.bak' WITH INIT,STATS;
GO

Let me explain what I am doing.

In lines 1-8 I create a  sample database Football.

In lines 10-24, I create a table, Footballer and a unique clustered index on column FootballerID.

In lines 26-34 I populate the Footballer table.

In lines 36-37 I take full database backup.

In lines 39-40 I take log database backup.

In lines 42-43 I rebuild the clustered index.

In lines 46-47 I take another log backup.

In lines 49-50 I rebuild the clustered index.

In lines 52-53 I take another full database backup. If a database backup clears the log, the next transaction log , should be small.

In lines 55-56 I take another log backup.

Let’s now compare the size of the last two transaction logs. Have a look at the picture below

As you can see the size of the transaction logs is more or less the same.So the last transactional log is not smaller,thus the full database backup did not clear the log.

The way I solved my friend’s problem was to free disk space so that the log can automatically grow and then take a full backup following a log backup.

Then I explained him the importance of log backups in log truncation and size management. He has incorporated log backups in his database backup strategy.

Hope it helps!!!!

I have been helping out a friend to understand various important issues regarding Transaction Logs in SQL Server databases.The transaction log records all the transactions and the modifications on the database those transactions have. Usually you have one transaction log file per database but you can have multiple files.The default size of a transaction log (if not specified) is the 25% of the total size of all data files created for the database.It is always a good idea to have the size of the transaction log planned beforehand.Bear in mind that the transaction log is zero initialized and that is a costly operation in both time and resources.The transaction log is divided into smaller parts that are called VLFs (Virtual Log Files). It is easier to manage the transaction log like that.  You have a active and inactive VLFs . One VLF must be active at any time in the database. You cannot configure the number and the size of the VLFs in the transaction log. As transaction log grows, new VLFs are created and their size is determined by SQL Server. There is a formula that determines the number and the size of the VLFs. If the transaction log grows by 64 MBytes, 4 new VLFs are created each 1/4 of the growth size.If the transaction log grows by 64 MBytes to 1Gbytes, 8 new VLFs are created each 1/8 of the growth size.If the transaction log grows by  1Gbytes or more, 16 new VLFs are created each 1/16 of the growth size.Also bear in mind that each VLF is uniquely identified by a sequence number.

Let’s have a look at the actual hands-on demos.

I will be using SQL Server 2012 Εnterprise  edition in this post but feel free to use SQL Server 2008/R2 versions and any other edition you might have installed (Standard, Express e.t.c).

1) Launch SSMS and connect to the local (or any other instance) of SQL Server

2) I will be looking firstly at the DBCC LOGINFO command which provides us with important information about the structure of the transaction log.

3) In a new query window in SSMS , type (or copy paste) the following t-sql script and then execute it

USE MASTER;
GO

CREATE DATABASE [Football] ON  PRIMARY
( NAME = N'Football_Data', FILENAME = N'D:\DATA\Football_Data.mdf' , SIZE = 180MB , MAXSIZE = UNLIMITED, FILEGROWTH = 20MB )
LOG ON
( NAME = N'Football_Log', FILENAME = N'D:\DATA\Football_log.ldf' , SIZE = 40MB , MAXSIZE = UNLIMITED, FILEGROWTH = 20MB)
GO

DBCC LOGINFO(N'Football');
GO

Let me explain what I do in this example. I create a database Football.Then I run the DBCC LOGINFO(N’Football’) to understand a bit about the log structure and the number of VLFs

Have a look at the photo below

As you can see there are 4 VLFs.Each one of them is 10 Mbytes.So we can see that the formula applies in this case. The Log file size is 40 Mbytes so we expect to have 4 VLFs.

4) Now let’s have a look at the DBCC SQLPERF(LOGSPACE) command. This command provides transaction log space usage statistics for all databases. Have a look

In a new query window type and execute the following T-Sql command.

USE MASTER;
GO

DBCC SQLPERF(LOGSPACE);
GO

Have a look at the picture below

5) Now I am going to increase the size of the transaction log by inserting sample data in the database and then check again the number and the size of the VLFs in the transaction log. I will also check the transaction log file size and the part of it that is used.

Type (copy and paste) the following T-SQL script. Execute it

USE FOOTBALL;
GO

ALTER DATABASE Football MODIFY FILE (NAME =N'Football_Log',SIZE=60MB);

DBCC LOGINFO(N'Football');
GO

DBCC SQLPERF(LOGSPACE);
GO

6) We add more space in the transaction log and then execute the DBCC LOGINFO and DBCC SQLPERF commands. Have a look at the picture below

We can see now that the size of the transaction log is more than 64Mbytes we have 8 VLFs

7) Now in a new query window type (copy paste) and execute the following script.

USE FOOTBALL;
GO

CREATE TABLE Footballer
(
FootballerID int identity(1,1),
Footballer_No varchar(20),
Footballer_lastname nvarchar(50),
Footballer_firstname nvarchar(50),
Footballer_middlename nvarchar(50)
)
GO

CREATE UNIQUE CLUSTERED INDEX cx_Foo_IX ON Footballer (FootballerID)
GO

declare @CTR INT
declare @ctrstr varchar(7)
select @ctr=0
WHILE @CTR<100.000
BEGIN
select @ctr=@ctr+1
select @ctrstr=Str(@ctr)
insert footballer(Footballer_No ,Footballer_lastname,Footballer_firstname, Footballer_middlename) values('F_No'+@ctrstr,'Footlname'+@ctrstr,'Footfname'+@ctrstr,'Footmname'+@ctrstr)
END

SELECT * FROM Footballer;
GO

I create a table in the database.Then I create a clustered index and insert 100.000 rows in it. That will increase the size of the used space in the transaction log.

Type and execute the T-SQL code below.

DBCC LOGINFO(N'Football');
GO


DBCC SQLPERF(LOGSPACE);
GO

Have a look at the picture below to see results I have got when I run the T-SQL code above.

Hope you learned about transaction log architecture and the importance of DBCC LOGINFO and DBCC SQLPERF commands.

Hope it helps!!!!

NopCommerce should certainly be in the toolbox of every .NET web company but as feature-rich and as extensible as it might be, you cannot offer the system to clients as it is, at least not with the default Dark Orange theme. But this is a normal part of the open source model and that is why every such system has a network of third party vendors to fill the gaps in the free software.

NopCommerce has quite a lot of customization partners and a few product vendors. But one particular set of products stand out from the rest.

In this blog post and the posts to follow I will try to show you how you can use the nopCommerce themes and extensions from Nop-Templates.com to build a solution for your e-commerce clients in literary minutes.

Looking at the default nopCommerce installation , http://demo.nopcommerce.com/,the first thing that I would like to have on my e-shop and probably the first thing that my e-commerce client will ask me for is how to put banners on the web site.  As a .NET web developer you have two options. Go out and find a sliding JQuery script, write some MVC code to integrate the JQuery script in wherever is appropriate in nopCommerce, and hard code some sliding images. This solution while possible will certainly require time researching and developing the code. If you are not familiar with nopCommerce you will have to factor in the learning curve too.

Fortunately we have the Nop Anywhere Sliders from Nop-Templates which will not only do all this but will also offer you a set of features, which will empower your nopCommerce with the possibility to put sliding images on any page and on as many pages as you need.

As a nopCommerce savvy user I have found working with the Nop Anywhere Sliders quite easy and intuitive. First of all you need to install the plugin via the nopCommerce administration:

It is important that the SevenSpikes.Core plugin is installed before the Nop Anywhere Sliders.

Once you have the Nop Anywhere Sliders installed you can go to the administration of the plugin by clicking on the Plugins -> Nop  Anywhere Sliders -> Manage Sliders:

A slider according to the Nop-Templates terminology is a set of images and settings, which are setup to slide in one place. You can define as many of these sliders as you need and add them to any part of your web site. For example you can define a slider called HomePageMainSlider:

Make sure that the slider dimensions are the same as the dimensions of your images. Most of the Slider Settings are self-explanatory and have a tooltip. If you need clarification on some of them you can visit the documentation page of the plugin which is also accessible via the Plugins -> Nop Anywhere Sliders -> Help Menu:

http://www.nop-templates.com/Help/NopAnywhereSliders/Version_2_7/lessons/Working_with_Nop_Anywhere_Sliders.html

Once you have configured your slider you can add some images to it by going to the Pictures tab:

For every image you can set the following settings:

When you have you slider and images set up, what is left to do is to add it to your web site. You have several ways of doing this. The easiest way is add it via the predefined widget zones:

If we add the slider to the home_page_slider_top widget zone, it will appear on our home page just below the main menu. You have more than 20 predefined widget zones by default which should be plenty to position your sliders wherever you need on your nopCommerce web site. However if you do not find an appropriate widget zone you can manually integrate the slider. To do this you need to add the following line of razor code on the page where you want you slider to appear:

@Html.Action(“Slider”, “AnywhereSliders”, new { systemName = “HomePageMainSlider”})

Alternatively you can define your own widget zones and add them to the plugin so that you can use them for your sliders. For more information on this you can read the documentation:

http://www.nop-templates.com/Help/NopAnywhereSliders/Version_2_7/lessons/Integration.html

One of the coolest features that the Nop Anywhere Sliders is the ability to map sliders to categories:

This is an important functionality for cases where you need a different slider for every category. The category page is the same for every category. It pulls the data for the category dynamically based on the id or name specified in the query string. Therefore you need this mapping so that the plugin knows when a specific category is being loaded which slider to add to the page. The category mapping is also supported by the manual integration of the plugin. If you are to add a slider manually to your category page you need to use the following line of Razor code:

@Html.Action(“CategorySlider”,”AnywhereSliders”, new { categoryId = Model.Id })

The category mapping of the plugin is quite a nifty feature, which can save you tons of coding and integration if you are to implement this from scratch.

I have covered the basics of working with the Nop Anywhere Sliders and I hope you will find this useful. The plugin is quite powerful and flexible and just does the job that it is supposed to do with very little effort required from the nopCommerce store owner or developer.

For more information do visit the product page:

http://www.nop-templates.com/p/4/nop-anywhere-sliders

I have installed SQL Server Developer edition in my machine. Developer edition of SQL Server 2012 can be installed in a client operating system.

1) I will use SQL Server Configuration Manager to change some initial configurations.We can manage the various services that are running and are related to SQL Server.We can manage the various network protocols.

Have a look at the picture below to see how you can open SQL Server Configuration Manager.

2) After firing up this tool, I can see all the SQL Server Services that are installed in this machine.You can see their names, their state, their start mode, that security account they are running under,the Process ID and the Service Type.You can see that there are 3 Database SQL Server Services running in my machine.One service for the named instance for the SQL Server 2008 R2 Enterprise Edition, one named instance for SQL Server 2012 Express Edition and one service for the default instance of SQL Server 2012 Developer Edition.

Have a look at the picture below

3) We can Start, Stop, Pause, Resume, Restart each service. Have a look at the picture below

4) We can also set/change the various properties by clicking on Properties.

We can change the Start Mode of a particular service e.g From Automatic to Manual.Have a look at the picture below.

5) We can also configure settings that are relevant with network protocols.These are going to be the protocols that SQL Server listens to for client connections.

When I select the protocols for the MSSQLSERVER I see there are 3 protocols:Shared Memory,Named Pipes,TCP/IP.If you want to allow remote connections from clients over a network like the Internet that supports TCP/IP we must enable that protocol.

Have a look at the picture below

We can also change some settings regarding the TCP/IP protocol by clicking on Properties.We can change the default  TCP/IP port number 1433, to something else.

After changing these settings you may need to restart the service.

Have a look at the picture below

6) Another thing you can do with SQL Server Configuration Manager is configuring client connections to other SQL Servers, meaning how this SQL Server instance is going to make outbound database connections.

Have a look at the picture below.Please take a note on the order of the protocols.You can see that all the protocols are enabled.

These are the protocols that SQL Server uses to connect to other instances of SQL Server. We can also set aliases from SQL Server Configuration Manager.A client will connect to the alias ( a name other the server instance name ).You can set the alias name, the port number,the protocol and the server it will connect to.

Have a look at the picture below

7) In order to show you the rest of SQL Server configurations I will switch to the familiar environment of SQL Server Management Studio – SSMS.Make sure you close the SQL Server Configuration Manager snap-in.

I will connect through SSMS to my SQL Server 2012 Developer Edition using Windows Authentication.Then I select the SQL Server instance and right-click on it. From the available options I choose Properties. In this new window I can set  various options regarding memory,processors and security.Do not make changes that affect the whole SQL Server instance unless you know exactly what you are doing.You should consult the product documentation. I will be going through most of these options in the next steps.In the General tab you can see information about the name of SQL Server instance, the language, the number of available processors,the amount of memory.

Have a look at the picture below

This information is read only.

8) We can also click on the Connection Properties in the Server Properties Window.
You can see the Authentication Mode and the username I used to log in.I am connected to the master database. You can also see the product name and version of SQL Server. There is also adequate information regarding the operating system.This information is also read only.

Have a look at the picture below

9) The next option in the Server Properties window is the Memory tab.

Have a look at the picture below

Most of the default options in this tab will be ok for most SQL Server configurations , unless you have heavy loads.Minimum and maximum server memory settings let you control how much memory SQL Server can use. We all know SQL Server is very demanding when it comes to memory.You should change these settings if you have another server application running in the same machine e.g Exchange Server. By the way you should never do that, having another server application running in the same machine where our production SQL Server lives. Some people ask me how much memory SQL Server is using.Well the answer to that should be “All of it”.

If you have 32 Gbytes of RAM in the box that SQL Server lives and there is only one database attached on this server that has a size of 32 Gbytes,SQL Server will cache the entire database in memory.This will improve performance significantly.If you add more memory to the server, make sure you will increase the Maximum server memory as well.

To recap the default memory settings and in particular the maximum server memory setting,if unchanged allows SQL Server to use as much memory as it wants. Well SQL Server listen to calls from the OS and it releases memory back to it so the operating system is not starved.You can change the settings from the Server Properties window.You can also use T-Sql to do that

Have a look at the picture below to see the statements I am using to set the maximum memory to 3800 Mbytes. In my machine the available Physical RAM is 4Gbytes.

Some people suggest that we should limit this number (maximum server memory) to 80-85% of the actual physical memory on a server that has only the Database Engine running.

You could use T-SQL to get the values for minimum and maximum memory settings.

In a new query window in SSMS, type and execute the following statement.

SELECT
    name,
    minimum,
    maximum,
    value_in_use
FROM master.sys.configurations
WHERE name IN (‘min server memory (MB)’, ‘max server memory (MB)’);

You can also use a DBCC command to get a snapshot of the current memory status of SQL Server.

In a new query window you could type DBCC MEMORYSTATUS. Have a look in this link for more information.

In the Memory tab  you could see another option if you are run a 32-bit server.This option that does not show up in my case since I am running a 64-bit version of SQL Server (as we saw earlier) is Use AWE to Allocate Memory .Do check this option for 32-bit servers with >4GB of memory, and keep it unchecked in all other cases.

Some of the changes I mentioned above will take effect immediately but in some cases you must restart the service.

10) The next set of options I will be looking into are the ones located in the Processors tab.

Have a look at the picture below

You can see that there are 4 available processors in this computer.SQL Server supports many processors and takes advantage of their power.

If you want to change the I/O Affinity or the Processor Affinity  you must first uncheck the two boxes under the label Enable processors.

Let me explain what I/O Affinity and Processor Affinity mean. By setting the Processor Affinity to one or more processors you basically dedicate SQL Server to use one or more of these processors.

By setting the I/O Affinity to a specific processor you basically bind I/O threads to this  processor.

I would advise against changing the values for I/O Affinity and Processor Affinity . Leave the default settings.

Another option I want to talk about is Boost SQL Server priority.BY enabling this option, SQL Server process will run as High Priority instead of the usual Normal priority. In most cases this will not bring any benefit to your database applications so leave the default value.

Another option I want to talk about is Maximum worker threads.This helps limit the resources consumed on an SQL Server instance by client connections. A new Windows thread is created for each connection.If you set a maximum it can hurt performance.Once that limit  is reached any new connection must wait for another connection to close in order to have access to the server.

By default, the max worker threads setting is 0 in SQL Server 2005,SQL Server 2008 and SQL Server 2012.Do not change the default value.

11) In the next tab we will look into settings regarding Security in SQL Server.I have already blogged about SQL Server Security in this post.

Have a look at the picture below.

We have two authentication modes in SQL Server,Windows Authentication Mode and Mixed Mode.

Windows Authentication basically means that a user must log in to a windows domain or a windows local account first before logging into an SQL Server system.Basically a security token is created by Windows when someone logs in and this is passed to SQL Server that “validates” the account  name and password based on that token. In a sentence the user logged in SQL Server is checked and identified by Windows.

SQL Server validates the account name and password using the Windows principal token in the operating system.

Windows Authentication disables SQL Server Authentication.I have been asked if you can disable Windows Authentication and the short answer to that is NO.

When possible (Microsoft recommends it) use Windows Authentication only since it is inherently more secure.For example when using Active Directory you can use Group Policies which means we can enforce certain policies like “enforce users to use strong passwords”.Another very useful policy is the “Account Lockout” which locks accounts when too many failed login attempts occur.

We also have Mixed Authentication Mode which basically means Windows Authentication & SQL Server Authentication.

With SQL Server Authentication mode SQL Server manages a seperate set of users and groups to give permissions to.SQL Server logins are not based on Windows accounts,usernames and passwords are created by SQL Server and also stored there.

If you choose to have SQL Server Authentication you would have to keep credentials for your windows domain and SQL Server.When using the SQL Server Authentication make sure you create a long sa password and hide it.

In any case you need to have this authentication mode enabled especially if you need to provide access to your users outside the organisation.SQL Server Authentication does not require a domain controller.

We can obviously change the authentication mode from SSMS but we can also type the following in a new query window

USE master
EXEC xp_instance_regwrite N'HKEY_LOCAL_MACHINE',
N'Software\Microsoft\MSSQLServer\MSSQLServer', N'LoginMode', REG_DWORD, 2

--Windows Authentication mode
USE master
EXEC xp_instance_regwrite N'HKEY_LOCAL_MACHINE',
N'Software\Microsoft\MSSQLServer\MSSQLServer', N'LoginMode', REG_DWORD, 1

In the Login auditing section we can select what kind of logins we will audit.

We can also use C2 Audit mode. It configures SQL Server to record attempts to access statements and objects.

C2 Audit mode can still be configured in SQL Server but is now rarely used. It has a negative performance impact through the generation of large volumes of event information.

When you set this option you should be prepared to run out of disk space.If this happens SQL Server will shut itself down.That is why I think it is imperative for everyone to check through jobs and alerts the remaining disk space where our databases live.

12) The next set of options I will be looking into are the ones located in the Connections tab.

Have a look at the picture below

You can set the maximum number of concurrent connections.That means if you set the number to anything other zero,then it is going to limit server resources that can service client connections.Be very careful with that setting.

You can also change the setting “Use query governor to prevent long-running queries”. This means that if we have long-running queries that use all the resources and put huge pressure on SQL Server resources.

We can limit the execution time of every query be checking this option. The default value is 300.That is 300 seconds by the way.That means that if the query does not complete within 300 seconds it will be aborted and will rollback the transaction.

You can also allow remote connections to the server.The default value is 600 seconds, that means 10 minutes.This option is enabled by default in SQL Server 2012 but not in previous versions. Do enable this option if you need it. Do not change the default value.

13) The next set of options I will be looking into are the ones located in the Database Settings tab.

Have a look at the picture below

The first option is the setting for the index fill factor.This specifies the amount of empty space that SQL Server leaves when an index is rebuild or created.This setting should be low (this is a percentage value) for databases that data does not change much and higher in OLTP databases where data is constantly updated,inserted or deleted.

We also have the option to use compressed backups.

Backup compression is available in the SQL Server 2008 R2 Standard edition . When we compress backups we save backup space and also reduce backup times.Obviously we will need more CPU resources to perform compression during a backup or decompression during a restore.One thing to keep in mind is that if we have databases that use data compression or are encrypted, backup compression will not be very efficient. Backup compression does not use row-level or page-level data compression. Backup compression happens at the time of a backup, and it uses its own compression algorithms.

We can also set the  “Default backup media retention” option.This option specifies the length of time to retain each backup set.

You can change this setting using T-SQL.

In a new query window type and execute the following statement

USE AdventureWorks2012 ;
GO
EXEC sp_configure 'show advanced options', 1;
GO
RECONFIGURE ;
GO
EXEC sp_configure 'media retention', 90 ;
GO
RECONFIGURE;
GO

Another option we can set is Recovery Interval. This option sets the maximum number of minutes for a database that Microsoft SQL Server needs to recover databases. The default value is 0, which lets SQL Server to automatically configure the recovery interval.

We can also change the Data Default locations option if we choose to.I do not see any reason why you want to change that.This is a setting that you change without fear that you will break something.

14) The next set of options I will be looking into are the ones located in the Advanced tab.

Have a look at the picture below

There are many options we can set in this tab. One of them is “Allow triggers to fire others“.This option is on by default. What actually means is that a trigger will allow to fire another trigger.Sometimes if we are not careful with that option we can get in an infinite loop. Another very important setting is the Optimize for Ad hoc Workloads.You can find more about this very useful option here.

Another option is Max Degree of Parallelism  and sets the number of processors used for the execution of a query with a parallel plan.

For best practices and recommendations on Max Degree of Parallelism have a look at this link.

Bear in mind that these are advanced settings and I would not touch them after the initial installation of SQL Server.

15) The next set of options I will be looking into are the ones located in the Permissions tab.

Have a look at the picture below

In this tab we see a list of principals and their permissions they have on the server level.I have selected the user fofo-pc\Nikos. This user is administrator in Windows. This user/principal is also in the fixed server role of sysadmin.You can see the permissions it has on the server.

Make sure you know exactly what each of the options I mentioned does and only then change them.

Hope it helps!!!

I was talking with SQL Server developers the other day and they have never heard or looked into the Resource Database.

The Resource database is available since the SQL Server 2005 version.

I assume that you have installed a version of SQL Server 2008/2012 in your machine.

I also assume that you are familiar with the SQL Server system databases – master,model,tempdb,msdb.

Most people have not heard about the Resource database because it is sort of “secret”,”hidden”.Some people look to find it in the SSMS. It won’t appear there.

You cannot backup the Resource database.Each SQL Server instance has one Resource database.

It is a read-only database that contains all SQL Server system objects.The actual objects are stored in the Resource database.These objects appear logically in the sys schema of each database attached in our SQL Server instance.

Have a look at the picture below. I have logged into my local SQL Server 2012 Developer Edition and expanded the System Views node of the AdventureWorks2012 database.

The Resource database makes upgrades to new versions of SQL Server easier, faster more efficient.The actual name of the database is mssqlsystemresource.mdf. It is located in the C:\Program Files\Microsoft SQL Server\MSSQL11.MSSQLSERVER\MSSQL\Binn path. It is the same path that SQL Server was installed in my machine. If you want to back up this database simply copy and paste it somewhere else using the Windows Explorer.

If we want to find the version of the Resource database, in a new query window type,

SELECT SERVERPROPERTY(‘ResourceVersion’) AS VersionOfResourceDatabase
GO

In my case the output is

VersionOfResourceDatabase
—————————————————————————————————————————————————————————————————————————————————————-
11.00.2845

As you see I am using the built in system function SERVERPROPERTY

If you want to find out when it was the last time the Resource database was updated you can type this bit of TSQL code in a query window in the SSMS.

SELECT SERVERPROPERTY(‘ResourceLastUpdateDateTime’) AS ResourceDatabaseLastUpdated
GO

Ιn my case the output is

ResourceDatabaseLastUpdated
—————————————————————————————————————————————————————————————————————————————————————-
2012-09-01 04:02:07.380

If you want to get the SQL definition for the sys.indexes object, type the following TSQL code in a new query in SSMS

SELECT OBJECT_DEFINITION(OBJECT_ID(‘sys.indexes’))AS Sys_Indexes_Definition
GO

The output is

CREATE VIEW sys.indexes AS
    SELECT i.id AS object_id,
        i.name AS name,
        i.indid AS index_id,
        i.type,
        n.name AS type_desc,
        sysconv(bit, i.status & 0×8) AS is_unique,            – IS_IND_UNIQUE
        isnull(ds.indepid, 1 – (i.status & 0×100)/0×100) AS dat

Hope it helps!!!

Recently I asked someone that gave me a call regarding an SQL Server issue, to have a look at the SQL Server Log and he just could not locate it. That is why I decided to write a post for this great tool.

I will show you how to access the SQL Server Log using various ways.

I will be using SQL Server 2012 Developer Edition but the Express edition will suffice .

I connect to the SQL Server instance using SSMS. I log in as sysadmin.In order to access the SQL Server Log I expand the Management node and then the SQL Server Logs node. One can see all the available logs.We have 7 logs in total. The Current and 6 Archived ones.These logs are for the SQL Service Service and not for (e.g) the Agent Service.The Current Log is the log that SQL Server is currently using.

Please have a look at the picture below.

We can use the Log File Viewer to view the contents of the current Log or any archived log.We can filter,search  the log.

Please have a look at the picture below

When you stop and start the SQL Server Service it archives all of the logs.Archive #6 means 6 times ago SQL Server was stopped.

I am going to stop and restart SQL Server Service.What happens now is that the oldest log is deleted and a new one is added. Archive #5 becomes Archive #6.

Please have a look at the picture below.

We can configure the number of archives that SQL Server will have.Select the SQL Server Logs node, right-click on it and select “Configure“. There you can set the maximum number of error log files.

Have a look at the picture below.

We can use the xp_ReadErrorLog stored procedure to read the contents of the SQL Server Log.

In a new query window I type

USE master
GO

EXEC xp_ReadErrorLog

Have a look at the picture below

We can access the SQL Server Log by going to the installation directory of SQL Server. Then I locate the Log folder. Then I open with some sort of  application (Notepad ++) the log I want.

Have a look at the picture below

We can use the Event Viewer to have a look at the contents of SQL Server Log.

Have a look at the picture below.

I am sure that you realise that the best way to view the SQL Server Log is the SQL Server Log File Viewer.

Hope it helps!!!

I will be using SQL Server 2012 Developer edition in this post but feel free to use SQL Server 2008/R2 versions.I will be using the AdventureWorks2012 database.You can download it from the Codeplex website.

There are times when the application code (high level code) includes ad-hoc queries/statements and there are no stored procedures in the code when it should.There are many cases when developers do not even use parameters. By that I mean that the developer code is written in a way that does not make use of execution plan reuse. Some people will say “Hey man, I do not care I have enough memory in my SQL Server box”.Well, SQL Server just loves memory and it is the one hardware resource that we must keep an eye on.

It is never a good argument to claim that there is “enough hard disk or memory in my system” so I do not care about fixing things…

Imagine that you have a reused plan in the plan cache and because of ad-hoc queries/workloads that plan is pushed out of the plan cache.That does not help the performance of the system.When you have many plans in the cache that are not reused at all that can lead to procedure cache bloat.In this case you have many entries (execution plans) in the plan cache that are used only one and if this happens far too often good plans (heavily reused plans) are forced out of the plan cache.

There is a setting that can be configured in SQL Server to approach such issues when we have lots of adhoc queries.It is called Optimize for Ad Hoc Workloads and you can enable it by using the system stored procedure sp_configure. By default this setting is off.When enabling this feature you must keep in mind that it will affect newly created plans and not the ones that are already placed in the plan cache.

Let me move on with my hands-on examples.

Connect to an instance of SQL Server. Type (copy paste) the code below.Do not execute anything yet.

USE AdventureWorks2012;
GO

IF OBJECT_ID ( 'Person.uspGetPersons', 'P' ) IS NOT NULL
DROP PROCEDURE Person.uspGetPersons;
GO

CREATE PROCEDURE Person.uspGetPersons
@LastName nvarchar(50),
@FirstName nvarchar(50)
AS

SET NOCOUNT ON;
SELECT FirstName, LastName, Title
FROM Person.Person
WHERE FirstName = @FirstName AND LastName = @LastName;
GO

DBCC FREEPROCCACHE

SELECT * FROM dbo.LookIntoPlanCache

EXEC Person.uspGetPersons N'Miller', N'Dylan'
GO
EXEC Person.uspGetPersons N'Duffy', N'Terry';
GO
EXEC person.uspgetpersons N'Brown', N'Kevin'
GO

SELECT * FROM dbo.LookIntoPlanCache

I execute line 1-2 to make AdventureWorks2012 the current database. Then I will create a simple stored procedure in lines 8-17.In lines 4-6 I just check to see if the stored procedure already exists on the system. Then in line 19 I clear the plan cache and use the LookIntoPlanCache view to see the contents of the plan cache.There will be no contents at this stage.

Then I execute the lines 23-28 .There are 3 calls to the stored procedure.Please note that I have typed the name of the procedure with 3 different ways.

Finally I execute the LookIntoPlanCache view to check the contents of plan cache once more.Have a look at the picture below to see the results I got.

The plan for the stored procedure is compiled, cached and reused.Having typed the stored procedure’s name in a different way does not matter .

In a new query window type (copy paste) the following t-sql code.Do not execute anything yet.

USE AdventureWorks2012
GO

DBCC FREEPROCCACHE

SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency]
GO
SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency];
GO

SELECT  *
FROM    [dbo].[LookIntoPlanCache]
GO

DBCC FREEPROCCACHE

SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency]
SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency];

SELECT  *
FROM    [dbo].[LookIntoPlanCache]
GO

Execute line 4.Once more I clear the contents of plan cache.Then I execute 2 simple SELECT statements.Please note that I in the second SELECT statement I have added a ‘;’.

Ηave a look at the picture below to see the results returned with the execution of the statements.

You can see that I get 2 execution plans for an identical query.The first thing to note here is that we should try and write T-SQL ad hoc queries in the same way. The second thing is that ad-hoc statements get different treatment compared to stored procedures.

Then in line 17 I clear the cache once more.Then I execute (line 19-22) the two SELECT statements in the same batch.Then I look into the contents of the plan cache using the LookIntoPlanCache view.

Have  a look in the picture below to see what my results are.

It is one entry this time in the plan cache but the plan size is twice the size.

Let’s see now how we can optimise SQL Server for ad-hoc workloads

In a new query window type (copy – paste the following).Do not execute anything yet.

USE AdventureWorks2012
GO

DBCC FREEPROCCACHE

SELECT  *
FROM    [AdventureWorks2012].[Production].[ProductModel]
GO

SELECT  *
FROM    [AdventureWorks2012].[Purchasing].[ShipMethod]
GO

SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency]
GO

SELECT  *
FROM    [dbo].[LookIntoPlanCache]
GO

sp_configure 'show advanced options', 1;
GO
RECONFIGURE
GO
sp_configure 'optimize for ad hoc workloads', 1;
GO
RECONFIGURE

SELECT  *
FROM    [AdventureWorks2012].[Production].[ProductModel]
GO

SELECT  *
FROM    [AdventureWorks2012].[Purchasing].[ShipMethod]
GO

SELECT  *
FROM    [AdventureWorks2012].[Sales].[Currency]
GO

SELECT  *
FROM    [dbo].[LookIntoPlanCache]
GO

I execute line 1-2 to make AdventureWorks2012 the current database.Then I clear the cache once more and execute three SELECT statements (lines 6-16) (ad-hoc queries).

Then in lines 18-20 I look into the plan cache once more.

Have a look at the picture below to see the results I have got.

Those results were the ones I expected. If most of your workload or even 25-40% percent of your workload is ad-hoc queries you should optimise SQL Server differently.

Execute lines 23-29. That will enable SQL Server to optimise execution plans for ad-hoc queries.Then execute the same 3 SELECT statements (lines 31-41) as before.Finally execute the view (lines 43-45) to look into the contents of plan cache once more.

Have a look at the picture below to see the results I have got.

Instead of a compiled plan I have a Compiled Plan Stub.The size that each of these stubs is 272 bytes. That size  is significantly less (150 times) than the size that the original compiled plans occupied. What will happen if I run again the 3 SELECT statements in lines 31-41? When I execute these lines again along with the view that looks the contents of the plan cache, I get the following results

As you can see now there are no more Compiled Plan Stubs for the 3 SELECT statements but compiled plans that take “full” size in the plan cache.So, when the same query is executed for a second time the full execution plan is generated.

To recap, make sure you write your code using stored procedures and parameterised queries. If you do not have access to the high level code and that code has been written by an “ad-hoc query” enthusiast make sure you optimise SQL Server accordingly,thus to optimise SQL Server for ad-hoc workloads.

Hope it helps!!!!

I will be using SQL Server 2012 Developer edition in this post but feel free to use SQL Server 2005,2008 versions.I will be using the AdventureWorks2012 database.You can download it from the Codeplex website.

Let me talk a bit about query optimisation and T-SQL. T-SQL is declarative by nature. When we write T-SQL statements in a query window in SSMS and execute them,we just say to SQL Server “I want these results back”. We do not provide any details on how the results will be returned.If there was nothing else between our T-SQL code and the SQL Server Database engine, we simply would not get any results back.Luckily for us there is a very important component, the Query Optimizer that generates an imperative plan. By saying imperative I mean detailed. This plan that is called execution plan is what is actually executed by the relational engine.The query optimiser will not look for a perfect plan.It is a cost-based optimiser that must find an efficient plan.The optimiser when deciding upon the execution plan will take in to consideration the type of operations,statistics (must always be up to date),indexes,hardware resources (number of CPUs ,available memory),SQL Server edition,number of active concurrent connections and query hints.

The following steps  take place upon query receival from SQL Server:

• Parser
• Algebriser
• Optimiser
• Execution plan generation – caching
• Query execution = execution plan execution

The Parser checks the an incoming query validating it for correct syntax.The query is terminated if a syntax error is detected.On validating a query the parser generates the parse tree.

The Algebriser takes the parse tree and resolves all the names of the different objects and data types.The output of all these verifications and resolutions is a binary set of data called a query processor tree.

Based on the query tree the Optimiser finds the optimal solution to execute the query.

Plans are then cached and reused so you do not have to create a new plan every time a query is executed.That will improve the performance significantly since execution plans are costly in both time and resources.So, the first time a query is executed the plan is created,compiled and placed in the procedure-plan cache. When there are subsequent requests for that query,the query plan will be used.

Connect to an instance of SQL Server. Type (copy paste) the code below and execute it. Make sure you include the execution plan ( CTRL+M )

USE [AdventureWorks2012]

GO

SELECT  Sales.SalesOrderDetail.OrderQty ,
Sales.SalesOrderDetail.SpecialOfferID ,
Sales.SalesOrderDetail.UnitPrice ,
Sales.SalesOrderDetail.UnitPriceDiscount ,
Sales.SalesOrderHeader.Status ,
Sales.SalesOrderHeader.AccountNumber
FROM    Sales.SalesOrderDetail
INNER JOIN Sales.SalesOrderHeader ON Sales.SalesOrderDetail.SalesOrderID = Sales.SalesOrderHeader.SalesOrderID
ORDER BY Sales.SalesOrderDetail.UnitPrice DESC

Ηave a look at the picture below to see the results returned with the execution of the statement above along with the execution plan.

I will not be looking into the various components of this execution plan and what they mean. You can hover over each of these components and get information like IO cost and CPU cost.

Most SQL Server developers and administrators I know, strongly believe that DMVs are the best feature that has been added to SQL Server since 2005.I feel exactly the same way.Please have a look at this post of mine that I talk about new DMVs and DMFs in SQL Server 2008 SP1 and SQL Server 2012.

Wouldn’t it be nice to be able to look into the contents of the cache?We can do that by using various DMVs.

In a new query window type the following.Do not execute anything yet.

USE AdventureWorks2012
GO
--DO NOT USE THIS COMMAND IN A PRODUCTION SYSTEM

DBCC FREEPROCCACHE

SELECT * FROM sys.dm_exec_cached_plans

SELECT * FROM Person.Person

SELECT * FROM sys.dm_exec_cached_plans

select * from sys.dm_exec_query_plan(0x060005003A0E56049009B9FF0100000001000000000000000000000000000000000000000000000000000000);

select * from sys.dm_exec_sql_text(0x060005003A0E56049009B9FF0100000001000000000000000000000000000000000000000000000000000000)

Execute line 1-5 first. I clear the contents of plan cache with the DBCC FREEPROCCACHE command.

Then (in line 7) I use the sys.dm_exec_cached_plans dynamic view that returns a row for each query plan that is cached by SQL Server.

Please have a look in this link for more information about the sys.dm_exec_cached_plans view.

At this time I do not expect to see anything in the plan cache.Then in line 9 I execute a simple SELECT statement..

An execution plan is generated. I run again the sys.dm_exec_cached_plans (line 11) view and I get back a row for this cached plan. I find out the relevant handle.

In my case is “0x060005003A0E56049029B9FF0100000001000000000000000000000000000000000000000000000000000000″

In your case it will be different so make sure you use yours instead.

Now I want to find the actual execution plan.We can use the sys.dm_exec_query_plan dynamic view.This view returns the plan in XML format for the batch specified by the plan handle.

Please have a look at this link for more information about the sys.dm_exec_query_plan view.I need to provide the handle as an input parameter.

I execute line 13 and I get the query plan in XML format for the sql query “Select * from Person.Person”

Then I want to find the SQL statement that created that plan.I will use the sys.dm_exec_sql_text dynamic view.

Please have a look at this link for more information about the sys.dm_exec_sql_text dynamic view.I need to provide the handle as an input parameter.

I execute the statement in line 15.The results I get is something like this

dbid    objectid    number    encrypted    text
5              NULL          NULL                0              SELECT * FROM Person.Person

So I managed to get the handle of the SQL query and then get the execution plan for this query and finally the text for the query itself.

Well that seems like a lot of work and I will demonstrate how to combine all these views into a custom-helper view.

In the same query window type (copy paste) and execute the following.

CREATE FUNCTION ExecPlanAndSqlText2 (@planhandle varbinary (MAX))
RETURNS TABLE
AS
RETURN SELECT EST.text,ECP.objtype,ECP.cacheobjtype,ECP.usecounts,ECP.size_in_bytes,EQP.query_plan FROM sys.dm_exec_sql_text(@planhandle) AS EST CROSS JOIN
sys.dm_exec_query_plan(@planhandle) AS EQP
JOIN sys.dm_exec_cached_plans AS ECP
ON ECP.plan_handle = @planhandle;
GO

SELECT * FROM ExecPlanAndSqlText(0x060005003A0E56049029B9FF0100000001000000000000000000000000000000000000000000000000000000);

I create a helper function first using all the dynamic management views and their relevant columns and then I pass as an input parameter the handle.

When I run the last SELECT statement I get the following results.

As you can see now I have all the information I need in one function.

In the same query window type (copy and paste) the code below.Execute the code

CREATE VIEW LookIntoPlanCache
AS
SELECT  EPAST.*
FROM    sys.dm_exec_cached_plans AS ECP
CROSS APPLY ExecPlanAndSqlText(ECP.plan_handle) AS EPAST

SELECT * FROM LookIntoPlanCache

I create a view (LookIntoPlanCache) that collects all the plan handles and passes them to our user defined function. Finally I execute a simple SELECT to see the contents of the view.

When I do that  I get the following results.

Finally I have the SQL query text,the execution plan and the times this plan was used in one view.

Make sure you use this view to look into the contents of your Plan Cache.It works for every system.

Hope it helps!!!!

The supported operating systems are Windows 7, Windows Server 2008, Windows Server 2008 R2.

There is a chance that the SBPA will fail during installation. You probably will have to download and install Microsoft Baseline Configuration Analyzer V2.0 first.

After I install the SBPA tool, I launch it (Admin Priveleges).

Have a look at the picture below

Then I click Start Scan.A new window appears.In this window I can specify the instance name of the SQL Server I want to scan and then choose what SQL Server components I want also to analyse. In my case I specified the SQL Server Engine & Setup of my SQL Server Developer 2012 version.

Have a look at the picture below

Then I click Start Scan once more.Then the scanning begins.Upon completion of the scanning by the SBPA 2.0, a report is generated and is categorized into Errors, Warnings and Compliants. Have a look at the picture below

You will be able to see different errors that exist on the SQL Server instance in relation with the rules configured in the Best Practices Analyzer.We can also click the detailed explanation for each error. In my case the error I decided to look into tempdb-wrong configuration error.There is a also a resolution for the error.

Have a look at the picture below

You can expand the Warning section of the report.In my case one of the warnings is that the AdventureWorks2012 database is configured with the Simple Recovery Model.I will have to decide depending on my disaster/recovery requirements if I need to change that setting or not.

Have a look at the picture below

To recap, this tool validates an SQL Server instance against a fixed set of built-in rules that can help us identify common installation and configuration issues.

Hope it helps!!!

Let’s see that with a hands-on example.

I am using the AdventureWorks2012 database.Make sure you visit the CodePlex website and download it.Then attach it (restore it) to your local SQL Server instance.

I will be using SQL Server 2012 Developer Edition but the Express edition will suffice .

I connect to the SQL Server instance.

Type (copy and paste the code below) the code below in a new query window in SSMS. Make sure you review it. At first glance there is nothing wrong.Try to execute the query.

USE AdventureWorks2012
GO

SELECT  [JobTitle] ,
[BirthDate] ,
[MaritalStatus] ,
[Gender] ,
[HireDate] ,
[VacationHours] ,
[SickLeaveHours] ,
( [VacationHours] + [SickLeaveHours] ) AS [TotalLeaveHours]
FROM    [AdventureWorks2012].[HumanResources].[Employee]
WHERE   [TotalLeaveHours] >=70

You will receive an error.

Invalid column name ‘TotalLeaveHours’.

SQL Server executes internally the SQL query with a different logic. It does not follow the order of the statements as they are depicted in the SQL Query.The order of execution is the following

• FROM
• ON
• JOIN
• WHERE
• GROUP BY
• HAVING
• SELECT
• DISTINCT
• ORDER BY
• TOP

You can see now why my query failed. SELECT is evaluated after the WHERE clause.

One way to visualize what I am explaining, please have  a look (download) this poster.

Let’s rewrite the previous statement.

Execute the code below

USE AdventureWorks2012
GO

SELECT  [JobTitle] ,
[BirthDate] ,
[MaritalStatus] ,
[Gender] ,
[HireDate] ,
[VacationHours] ,
[SickLeaveHours] ,
( [VacationHours] + [SickLeaveHours] ) AS [TotalLeaveHours]
FROM    [AdventureWorks2012].[HumanResources].[Employee]
ORDER BY [TotalLeaveHours] DESC

You can see now why my query succeeds. ORDER BY is evaluated after the SELECT clause.

Another excellent article on the subject is this one (greek audience only :P).

I am big fan of LINQ to SQL and Entity Framework. I am sure you have written at least one query using the LINQ syntax.Have a look at the ( snippet )code below

var companies = from comp in ctx.Companies
                                   where comp.CompanyName.StartsWith(“Ad”)
                                   orderby comp.CompanyName
                                   select new { comp.CompanyID, comp.CompanyName };

I am sure you can see the similarities between LINQ query syntax and how SQL Server processes a query internally. So if you think that LINQ syntax is “strange” you know now why we write the LINQ queries the way we do.
Hope it helps!!!

I am using the AdventureWorks2012 database.Make sure you visit the CodePlex website and download it.Then attach it (restore it) to your local SQL Server instance.

I will be using SQL Server 2012 Developer Edition but the Express edition will suffice .
I connect to the SQL Server instance.

Type (copy and paste but make sure you review it) the code below.Execute the code below

USE [AdventureWorks2012]

GO

SELECT  [Title] ,
[FirstName] ,
[MiddleName] ,
[LastName] ,
[Suffix]
FROM    [AdventureWorks2012].[Person].[Person]
ORDER BY LastName ,
FirstName
OFFSET 0 ROWS
FETCH NEXT 10 ROWS ONLY;

In this first example I simply instruct SQL Server to skip 0 rows and then fetch me the next 10 rows.

In the same (or in a new query window ) type (copy and paste but make sure you review it) the code below.Execute the code below

SELECT  [Title] ,
[FirstName] ,
[MiddleName] ,
[LastName] ,
[Suffix]
FROM    [AdventureWorks2012].[Person].[Person]
ORDER BY LastName ,
FirstName
OFFSET 500 ROWS
FETCH NEXT 100 ROWS ONLY;

In this second example I simply instruct SQL Server to skip 500 rows and then fetch me the next 100 rows.

I will rewrite the code below to incorporate variables. In this case I simply instruct SQL Server to skip 1000 rows and then fetch me the next 1000 rows.

In the same query window type (copy-paste) but do not execute the code below.

DECLARE @offset INT = 1000;
DECLARE @rows_tofetch INT = 1000;

SELECT  [Title] ,
[FirstName] ,
[MiddleName] ,
[LastName] ,
[Suffix]
FROM    [AdventureWorks2012].[Person].[Person]
ORDER BY LastName ,
FirstName
OFFSET @offset ROWS
FETCH NEXT
@rows_tofetch ROWS ONLY;

In order to get the next page (skip 1000 rows and return 1000 rows after that) we need to rewrite the code above.

Review the code below and then copy and paste it in a new query window.

DECLARE @offset INT = 1000;
DECLARE @rows_tofetch INT = 1000;

SET @offset = 2 * @offset + @rows_tofetch

SELECT  [Title] ,
[FirstName] ,
[MiddleName] ,
[LastName] ,
[Suffix]
FROM    [AdventureWorks2012].[Person].[Person]
ORDER BY LastName ,
FirstName
OFFSET @offset ROWS
FETCH NEXT
@rows_tofetch ROWS ONLY;

Execute the code above.

Hope it helps!!!

If you use the sp_recompile system stored procedure before the actual stored procedure the SQL Server Optimiser will drop the existing plan from the procedure cache forcing a new plan to be created the next time that the procedure will run.You can also add the WITH RECOMPILE directly to the stored procedure declaration right before the AS keyword. This tells SQL Server to throw out the query plan on every execution of the stored procedure.

In this post I will be looking into how schema changes in objects that are referenced by the stored procedure, can cause recompilation.

I will be using SQL Server 2012 Developer edition in this post but feel free to use SQL Server 2005,2008 versions.I will also use SQL Server Profiler.

We all know that SQL Server optimises the execution of a sql query using execution plans.The execution plan is what is compiled and executed by the relational engine and then placed for reuse in the plan cache.T-SQL is declarative by nature.That means that it does not provide any details on how the results will be returned.On the other hand SQL Server execution plans provide the actual (imperative-detailed) way for the relational engine to fetch the results.One way of using the plan cache in an efficient way is exactly writing you T-SQL commands in the context of a stored procedure.

Let me demonstrate recompilation by creating a small example. I am using the AdventureWorks2012 database.

I connect to the SQL Server instance.

Type (copy and paste but make sure you review it) the code below.Make sure you do not execute anything yet.

USE [AdventureWorks2012]

GO

IF ( SELECT OBJECT_ID('dbo.GetProductInfoByListPrice')
) IS NOT NULL

DROP PROCEDURE dbo.GetProductInfoByListPrice

GO

CREATE PROCEDURE GetProductInfoByListPrice
@LowerVal MONEY,
@UpperVal MONEY
AS
BEGIN

SELECT  Name,
ProductNumber,
Color,Size,StandardCost
FROM    Production.Product
WHERE   [ListPrice] > @LowerVal
AND [ListPrice] < @UpperVal

END

GO

DBCC FREEPROCCACHE

SELECT cpl.objtype AS PlanType,
OBJECT_NAME(stl.objectid,stl.dbid) AS ObjectName,
cpl.refcounts AS ReferenceCounts,
cpl.usecounts AS UseCounts,
stl.text AS SQLBatch,
qpl.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cpl
CROSS APPLY sys.dm_exec_query_plan(cpl.plan_handle) AS qpl
CROSS APPLY sys.dm_exec_sql_text(cpl.plan_handle) AS stl;
GO

Execute lines 1-3 to bring AdventureWorks2012 to scope. Then in lines 5-10 I check to see if the stored procedure already exists. Then I create a stored procedure (lines 12-23) that returns a few columns from the Production.Product table based on two input parameters based on the ListPrice column values.In line 29 I use the DBCC command to clear the plan cache.Please do not execute this command in production systems.

In lines 31-40 I use 1 DMV and 2 DMFs to get information from the procedure cache.I get information such as the Plan Type,Use counts(how many times the plan was used),the actual query and its query plan.When you execute the code you will see that there are no plans in the plan cache.

Now we need to launch SQL Server Profiler. Go to the menu Tools -> SQL Server Profiler.The SQL Server Profiler will launch and I will connect using Windows Authentication.I will create a new trace based on the Blank template.In the events tab I will use the SP:Recompile and SQL:StmtRecompile.The SQL:StmtRecompile event fires every time a statement recompilation occurs during execution.The SP:Recompile event exists but only for backwards compatibility in SQL Server 2005.Make sure you hit the Run button.

Have a look at the picture below

In the same query window type (copy-paste) but do not execute the code below.

EXEC GetProductInfoByListPrice 7,155

EXEC GetProductInfoByListPrice 24,3456

SELECT cpl.objtype AS PlanType,
OBJECT_NAME(stl.objectid,stl.dbid) AS ObjectName,
cpl.refcounts AS ReferenceCounts,
cpl.usecounts AS UseCounts,
stl.text AS SQLBatch,
qpl.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cpl
CROSS APPLY sys.dm_exec_query_plan(cpl.plan_handle) AS qpl
CROSS APPLY sys.dm_exec_sql_text(cpl.plan_handle) AS stl;
GO

Execute lines 1-3.We execute the stored procedure with two different values. Then have a look again in the procedure cache (lines 5-14).

Ηave a look at the picture below to see the results I have got in my case.The execution plan was created, placed in the cache and reused.There is no activity in the SQL Server Profiler.

In the same query window type the following

CREATE INDEX IX_Test ON  Production.Product(ListPrice)

EXEC GetProductInfoByListPrice 55,5555

I am creating a new index on the Production.Product table and then I execute the stored procedure again. SQL Server Optimiser detects the presence of the new index and does not use the current execution plan in the procedure cache.It does recompile the existing plan because it is more beneficial to spend more CPU cycles to create a better execution plan.

Have a look in the picture below

To recap we saw that the SQL Server Optimiser will drop a plan from the procedure cache when it decides that are changes in the underlying objects that are referenced from the stored procedure that result in a more optimal plan.

Bear in mind that if you have a high recompilation ratio this is one factor that will slow your whole application.You must identify the reasons behind that and eliminate them.It costs both in terms of time and resources for SQL Server to create a new plan.

Hope it helps!!!

Well, most of you know that we have TRY-CATCH statements for exception handling since SQL Server 2005.I have blogged about error handling and exceptions in this post.

I am sure that most of you are familiar with TRY-CATCH statement but before I go on with the actual example I will explain it. Inside the TRY block all the instructions that might cause an exception are included. When an exception occurs from one of the statements that are inside in the TRY block, the control is handed to the CATCH block, where the exception can be handled (by printing a message to the user, log the error e.t.c)

In SQL Server 2012 we have a new statement, THROW which can be used as an alternative of RAISERROR but most importantly we can re-throw the original error that occurred back to the client, so it can be handled at the application level.

You can use any edition of SQL Server of 2012 you want for this example. SQL Server Express edition will do just fine.Download the Express edition, here.

Connect to the instance of the SQL Server.

Before I move on to demonstrate the use of THROW statement, there are some things we must note

• The line (statement) before the THROW statement must end with semicolon (;)
• All exceptions raised by THROW statement will have a severity of 16
• THROW statement doesn’t require an error number but if you choose to use one it should be greater than 50.000

The syntax is pretty basic:

THROW (error_number,message,state)

You must define these parameters only if you use THROW outside the CATCH block.

1) THROW outside a catch block

In a new query window type and execute the code below

THROW 12767, 'I am a new statement in SQL Server 2012', 1;
THROW 52767, 'I am a new statement in SQL Server 2012', 1;

The first statement will fail since the error number is less than 50.000. The second statement will succeed.

2) THROW inside a catch block

In a new query window type  the code below

USE tempdb;
go

CREATE TABLE Persons
(
P_Id int NOT NULL,
LastName varchar(255) NOT NULL,
FirstName varchar(255),
Address varchar(255),
City varchar(255),
PRIMARY KEY (P_Id)
)

INSERT INTO persons VALUES (1,'Michael','Owen','122 liverpool street','liverpool')
INSERT INTO persons VALUES (2,'Steven','Gerrard','123 liverpool street','liverpool')

SELECT * FROM persons

CREATE TABLE Orders
(
O_Id int NOT NULL PRIMARY KEY,
OrderNo int NOT NULL,
P_Id int FOREIGN KEY REFERENCES Persons(P_Id)
)

BEGIN TRY
BEGIN TRANSACTION
INSERT INTO orders VALUES (2,12345,2)
INSERT INTO orders VALUES (4,12345,3)
COMMIT TRANSACTION
END TRY
BEGIN CATCH
RAISERROR ('we have an error',17,21);
IF @@TRANCOUNT > 0
ROLLBACK TRANSACTION
END CATCH

I create a table(Persons) in tempdb and then I insert some values in it.Then I create a second table that is related to the first one with a foreign key constraint.

Then I use an insert statement for the second table (Orders -That I know will violate the foreign key constraint) inside a TRY-CATCH block and use RAISERROR function to raise a custom error and then rollback the transaction.

Execute the code and note the results and messages.

Now let’s re-write the INSERT statement above using the THROW statement.

In a new query window type and execute the code below

USE tempdb;
go

BEGIN TRY
BEGIN TRANSACTION
INSERT INTO orders VALUES (2,12345,2)
INSERT INTO orders VALUES (4,12345,3)
COMMIT TRANSACTION
END TRY
BEGIN CATCH
PRINT 'There was an error when inserting the records';
THROW;
END CATCH

In my case the results are

The actual error that caused the exception in the first place, appears (is re-thrown) again.

Please note that there is not (yet) FINALLY statement.

Hope it helps!!!

DMO’s were first introduced in SQL Server 2005 and were immediately very popular with developers and administrators.They give us a great insight on what is going on inside  SQL Server.It would be very difficult otherwise to get  all this information so easily. They have great information on sessions,plan cache,memory,threads,database connections,waits,latches e.t.c.We can get all this information without placing a heavy burden on the server.

All of them start with dm_  in the name.They are all stored in the sys schema.

You can use any edition of SQL Server of 2012 you want for this example. SQL Server Express edition will do just fine.Download the Express edition, here.

Connect to the instance of the SQL Server.

The first DMF I will be looking into is sys.dm_os_volume_stats. This function returns several important information about the physical disk drive the specified database and its respective files are stored in SQL Server 2012.It returns information such as: Available free space on the volume,mount point at which the volume is rooted,whether the volume supports sparse files or not,whether the volume is currently marked as read only or not,whether the volume is currently compressed or not, e.t.c

As input parameters we need to provide the Database Id and the File Id values.

In an new query window type and execute the following code.Note the results.

USE AdventureWorks2012;
GO

SELECT name,total_bytes,available_bytes,volume_mount_point,file_system_type,is_compressed
FROM sys.database_files df
CROSS APPLY sys.dm_os_volume_stats(db_id(),df.file_id)

The next DMV is sys.dm_os_windows_info.It information about Windows Operating System.Type and execute the code below. Note the results.

USE MASTER;
GO
SELECT * FROM sys.dm_os_windows_info;

The next one I want to talk about is (sys.dm_server_services) and returns the list of SQL Server’s windows services.Type and execute the code below.Note the results.

USE MASTER;
GO
SELECT * FROM sys.dm_server_services

The next one I want to talk about is (sys.dm_server_memory_dumps) and returns  information about each memory dump file generated by SQL Server.

Type and execute the code below.Note the results.

USE MASTER;
GO
SELECT * FROM sys.dm_server_memory_dumps;

The next one I want to talk about (sys.dm_server_registry) and returns configuration and installation information stored in the registry for the current SQL Server instance.

Type and execute the code below.Note the results.

USE MASTER;
GO
SELECT * FROM sys.dm_server_registry;

The next one I want to talk about  is  (sys.dm_exec_query_stats). This is not a new one but it has changed in this version of SQL Server. 4 new columns have been added.

•     total_rows –  This column returns the total number of rows returned by the query

•     last_rows –    This column returns the total number of rows returned by last execution of the query

•     min_rows –   This column returns the minimum number of rows query returned over the number of times that the plan has executed since it was last compiled

•     max_rows –  This column returns the maximum number of rows query returned over the number of times that the plan has executed since it was last compiled

I will be using AdventureWorks2012 for this example. I run the first query below. I note that it returns 655 rows.

Then in the second query, I am returning old and new columns from the sys.dm_exec_query_stats for the query where the total number of rows is 655.

Type and execute the code below.

USE AdventureWorks2012;
GO

SELECT        Sales.SalesOrderHeader.OrderDate, Sales.SalesOrderHeader.DueDate, Sales.SalesOrderHeader.ShipDate, Sales.SalesOrderHeader.SubTotal,
Sales.SalesOrderDetail.UnitPrice, Sales.SalesOrderDetail.OrderQty
FROM          Sales.SalesOrderHeader INNER JOIN
Sales.SalesOrderDetail ON Sales.SalesOrderHeader.SalesOrderID = Sales.SalesOrderDetail.SalesOrderID
WHERE Sales.SalesOrderHeader.ShipDate > '2008-07-29'
ORDER by UnitPrice

SELECT execution_count,total_logical_reads,total_logical_writes,total_physical_reads,total_worker_time,total_rows,last_rows,min_rows,max_rows
FROM sys.dm_exec_query_stats
WHERE total_rows = 655

There are many great posts on the web about DMVs but one post I really like is Glen Berry’s , A DMV per day series.

Hope it helps!!!!

In this post I will also show you how to corrupt a database by interfering (corrupting) with some database pages.

The Restore Pages feature let us restore damaged pages without restoring the entire database.These pages are marked as suspect.

Restore pages is only possible in the full and bulk-logged recovery model.We can only restore database pages.Page Restore must start with a full, file, or filegroup backup and there should be no break in the chain of log backups.

You can use any edition of SQL Server of 2012 you want for this example. SQL Server Express edition will do just fine.Download the Express edition, here .

In my case I will use Developer edition of 2012 for these example.

I connect to the SQL Server instance.

Type (copy – paste) the code below in a new query window

USE MASTER;
GO

CREATE DATABASE FOOTBALL;

USE FOOTBALL;
GO

CREATE TABLE Footballer
(
FootballerID int identity(1,1),
Footballer_No varchar(20),
Footballer_lastname nvarchar(50),
Footballer_firstname nvarchar(50),
Footballer_middlename nvarchar(50)
)

CREATE UNIQUE CLUSTERED INDEX cx_Foo_IX ON Footballer (FootballerID)

declare @CTR INT
declare @ctrstr varchar(7)
select @ctr=0
WHILE @CTR<60000
BEGIN
select @ctr=@ctr+1
select @ctrstr=Str(@ctr)
insert footballer(Footballer_No ,Footballer_lastname,Footballer_firstname, Footballer_middlename) values('F_No'+@ctrstr,'Footlname'+@ctrstr,'Footfname'+@ctrstr,'Footmname'+@ctrstr)
END

SELECT * FROM footballer

BACKUP DATABASE [FOOTBALL] TO  DISK = N'C:\Program Files\Microsoft SQL Server\MSSQL11.SQL2012\MSSQL\Backup\FOOTBALL.bak'
WITH NOFORMAT, NOINIT,  NAME = N'FOOTBALL-Full Database Backup', SKIP, NOREWIND, NOUNLOAD,  STATS = 10
GO

I am creating a database that I will corrupt in the process.Then I create a table and clustered index on that table.Then I populate the table with 60.000 records.

Then I take full database backup of the database.This backup contains un-damaged pages.

I would like to corrupt a page that is part of the clustered index.I will use the DBCC IND command.DBCC IND command provides the list of pages used by the table or index.

DBCC IND ('FOOTBALL', 'footballer',1)

Have a look at the picture below to see the results and the page I selected (1267).

Before I move on with the corruption of that particular page I must take the database offline.

Type and execute the code below

USE master;
ALTER DATABASE FOOTBALL SET OFFLINE;

Now I need to calculate the starting offset of the page I am about to corrupt.

SELECT 1267*8192 AS [StartingOffset]

In my case is 10379264

Now we need to open the .mdf file in a hexadecimal editor. You can download this free Hex editor here

It is very easy to install and run the editor. Make sure you launch this application with elevated privileges.Then navigate to the location of the .mdf file and open it.

Inside this file we need to locate our page. I press  GoTo (or Ctrl + G), then paste in the offset value (10379264) by selecting decimal.Then I press OK  and I am at the beginning of this page.

Have a look at the picture below

Now I am about to corrupt by simply changing the values and corrupting the page.

Have a look at the picture below

Now we need to bring the database online.

Type and execute the code below

USE master;
ALTER DATABASE FOOTBALL SET ONLINE;

Then we can run a simple query

Type and execute the code below

USE FOOTBALL;
SELECT FootballerID FROM Footballer

We will get an error. Have a look at the picture below

Now let’s run the DBCC CHECKDB command.

DBCC CHECKDB ('FOOTBALL')

Have a look at the picture below to see the results I have got.

You can query the suspect_pages table of  the msdb database.

SELECT * FROM msdb..suspect_pages

In my case this is the output I get when I run the statement above.

database_id    file_id     page_id              event_type       error_count            last_update_date
———– ———– ——————– ———– ———– ———————–
6                            1           1267                            1                             3                  2012-07-18 03:17:57.333

You can also have a look at the SQL Server Logs.You will identify the corruption immediately.

Have a look at the picture below

In order to restore the page, I select the database, I right-click on it (from the Object Explorer in the SSMS) then select Tasks->Restore->Page .

Have a look at the picture below

Then the Restore Page screen appears.You can see that the database is selected, the FileID and PageID of the damaged page.You can run DBCC CHECKDB command.You can also see the Tail-Log backup and all the other backups.
Have a look at the picture below

Then I press OK.Then I execute the statements below and there are no errors.

USE FOOTBALL;
SELECT FootballerID FROM Footballer
DBCC CHECKDB ('FOOTBALL');

To recap, we showed (do not do this in a production database) how to corrupt a page in a database and also how to perform Page Restore using SSMS UI in SQL Server 2012.

Hope it helps!!!

Many people think that the execution plan of a stored procedure is cached in the procedure cache when the stored procedure is created. That is a mistake.

You can use any edition of SQL Server of 2012/2008/2005 you want for this example. SQL Server Express edition will do just fine.Download the Express edition, here .

In my case I will use Developer edition of 2012 for these example.

I connect to the SQL Server instance.

Type (copy and paste but make sure you review it) and execute the code below

USE AdventureWorks2012
GO

DBCC FREEPROCCACHE
GO

CREATE PROCEDURE PeopleinSeattle
AS
SELECT [AddressID]
,[AddressLine1]
,[AddressLine2]
,[City]
,[StateProvinceID]
,[PostalCode]
FROM [AdventureWorks2012].[Person].[Address]
WHERE [City] = 'Seattle'

SELECT cpl.objtype AS PlanType,
OBJECT_NAME(stl.objectid,stl.dbid) AS ObjectName,
cpl.refcounts AS ReferenceCounts,
cpl.usecounts AS UseCounts,
stl.text AS SQLBatch,
qpl.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cpl
CROSS APPLY sys.dm_exec_query_plan(cpl.plan_handle) AS qpl
CROSS APPLY sys.dm_exec_sql_text(cpl.plan_handle) AS stl;
GO

Let me explain what I do with the code above. I clear the procedure cache in line 4.Do not do that in a production environment.

Then in lines 8-17 I create a stored procedure.In lines 19-28 I use 1 DMV and 2 DMFs to get information from the procedure cache.

I get information such as the Plan Type,Use counts(how many times the plan was used),the actual query and its query plan.

When I execute lines 8-17 I get the result you see in the picture below.

It is clear that there is no stored procedure in the cache.Let’s execute the stored procedure and run the code with the DMO (DMV+DMFs) again

exec PeopleinSeattle

SELECT cpl.objtype AS PlanType,
OBJECT_NAME(stl.objectid,stl.dbid) AS ObjectName,
cpl.refcounts AS ReferenceCounts,
cpl.usecounts AS UseCounts,
stl.text AS SQLBatch,
qpl.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cpl
CROSS APPLY sys.dm_exec_query_plan(cpl.plan_handle) AS qpl
CROSS APPLY sys.dm_exec_sql_text(cpl.plan_handle) AS stl;
GO

The result follows in the picture below

Only when the stored procedure is executed the Query optimiser generates a plan, compiles it and places it in the procedure cache.No plan is generated when the stored procedure is created.

Execute the code below

exec PeopleinSeattle
GO 40

SELECT cpl.objtype AS PlanType,
OBJECT_NAME(stl.objectid,stl.dbid) AS ObjectName,
cpl.refcounts AS ReferenceCounts,
cpl.usecounts AS UseCounts,
stl.text AS SQLBatch,
qpl.query_plan AS QueryPlan
FROM sys.dm_exec_cached_plans AS cpl
CROSS APPLY sys.dm_exec_query_plan(cpl.plan_handle) AS qpl
CROSS APPLY sys.dm_exec_sql_text(cpl.plan_handle) AS stl;
GO

We execute the stored procedure 40 times and we check again to see the contents of the procedure cache.

Have a look at the picture below to see the results

Now we can see that the compiled plan that was compiled and placed in the procedure cache for this particular stored procedure will be used every time this procedure is executed.That is great performance boost.

That is one of the main advantages of stored procedures.

Hope it helps!!!

Obviously we can use the functions in the SELECT part of the query. In that case the performance impact is not that great.

You can use any edition of SQL Server of 2012 you want for this example. SQL Server Express edition will do just fine.Download the Express edition, here .

In my case I will use Developer edition of 2012 for these examples.

I connect to the SQL Server instance.

I need to create a sample database (SAMPLEDEMO) and a table (SALES) in it.I create a non-clustered index on the OrderDate column.

Type (copy and paste but make sure you review it) and execute the code below

USE [master]
GO

CREATE DATABASE [SAMPLEDEMO]

GO

USE [SAMPLEDEMO]
GO

CREATE TABLE [dbo].[SaleS](
[SalesOrderID] [int] IDENTITY(1,1) NOT FOR REPLICATION NOT NULL,
[OrderDate] [datetime] NOT NULL,
[DueDate] [datetime] NOT NULL,
[ShipDate] [datetime] NULL,
[Status] [tinyint] NOT NULL,
[SubTotal] [money] NOT NULL,
[TaxAmt] [money] NOT NULL,
[Freight] [money] NOT NULL,
[Comment] [nvarchar](128) NULL,
[rowguid] [uniqueidentifier] ROWGUIDCOL  NOT NULL,
[ModifiedDate] [datetime] NOT NULL,
CONSTRAINT [PK_SalesOrderHeader_SalesOrderID] PRIMARY KEY CLUSTERED
(
[SalesOrderID] ASC
)WITH (PAD_INDEX = OFF, STATISTICS_NORECOMPUTE = OFF, IGNORE_DUP_KEY = OFF, ALLOW_ROW_LOCKS = ON, ALLOW_PAGE_LOCKS = ON) ON [PRIMARY]
) ON [PRIMARY]

 

CREATE NONCLUSTERED INDEX [IX_Sales_OrderDate] ON [dbo].[SaleS]
(
[OrderDate] ASC
)

GO

Then I will populate the table with 10.000 rows of data using a third party tool (Data Generator from RedGate).

Now we will write 2 queries (we were asked  to produce a report from our e.g sales manager) that will bring back only the SalesOrderID and OrderDate values  for the orders that were placed on July 2008.

Type and execute the code below.Make sure that you have clicked on the “Include Actual Execution Plan” button.We need the metrics from the execution plan.

USE [SAMPLEDEMO]
GO

SELECT [SalesOrderID],[OrderDate]
FROM [SAMPLEDEMO].[dbo].[SaleS]
WHERE year(orderdate)= '2008' and MONTH(OrderDate) = '7'

SELECT [SalesOrderID],[OrderDate]
FROM [SAMPLEDEMO].[dbo].[SaleS]
WHERE orderdate between '2008/07/01' and  '2008/08/01'

Have a look at the picture below to see my findings

In both cases I get back 102 rows of data. In the first sql query using the built-in functions the cost is 91% relative to the batch and I see that I have an Index scan. By rewriting the query (omitting the built-in functions) I see that the cost is 9% relative to the batch and the optimiser uses an Index Seek which is a far better option.The non-clustered index we created is used in the second case because the optimiser can see now that the values in the OrderDate column are organised according to this index.This is a “range” query and because there is an index that is available on this column it is used efficiently to retrieve the requested data.

Hope it helps!!!!

You can download these reports here. The installation process is straight forward and you just need to follow the steps of the installer.

Once these reports (these are simply Reporting Services report files) have been installed we can use them through SSMS (Custom Reports).

There are Standard Server reports available to us.Let’s have a look at what standard reports are available to us at the server level.We have reports on queries performance (filtered by IO,CPU), transactions that cause blocking,locking,memory consumption on the server e.t.c.

Please have a look at the picture below

I have chosen the first report (Server Dashboard) – This report provides overview data about the SQL Server instance and its configurations.

Have a look at the picture below

There are Standard reports at the database level as well.We can have Disk Usage reports,backup and restore events reports and much more.

Have a look at the picture below

For this example I have chosen the Index Usage Statistics report.

Have a look at the picture below

Let me show you how to use SQL Server 2012 Performance Dashboard Reports after you install them. You just right- click on the server in the Object Explorer, then choose Reports from the context menu and then Custom Reports.

Have a look at the picture below

Then in the Open File Dialog box choose performance_dashboard_main report file (you must know the folder where all these reports are kept- e.g “C:\Program Files (x86)\Microsoft SQL Server\110\Tools\Performance Dashboard” folder)  and click Open to display the report.

Please have a look at the picture below that shows the output of this particular report.

Have a look at the rest of the Custom Reports (Waits,IO Statistics) by selecting the links from the main report.

Hope it helps!!!

This is a codeplex project and you can download it here . This tool helps to identify large objects,objects with a high index/data ratio and tables which may be growing more quickly than expected.

Installation of this tool is really straight forward. When you run it for the first time, you need to specify the server instance you are connecting to, the database and whether you are connecting using SQL Server or Windows authentication.

I launched the SQL Space Map application , I have connected to my local instance of SQL Server 2012 Developer edition through Windows authentication to the AdventureWorks2012 database. When I do that I can see a visual representation of the database objects in my database and their respective size. Make sure you hover over the various tables and indexes.

Please have a look at the picture below

We have also the Standard Reports that are available out of the box for SQL Server 2012 and provide similar information.But I prefer the visual representation that SQL Space Map tool provides.

Have a look at the picture below

Hope it helps!!!

In this rather sort post I will demonstrate with  hands-on examples how to use the new string function,CONCAT().

You can have a look in another blog post of mine related to built-in SQL functions.You can have a look into the new logical functions introduced in SQL Server 2012, here .

You can have a look in another blog post of mine explaining the new conversion functions in SQL Server 2012.

You can use any edition of SQL Server of 2012 you want. SQL Server Express edition will do just fine.Download the Express edition, here .

In my case I will use Developer edition of 2012 for these examples.

This function (CONCAT()) allows multiple string values to be joined (concatenated) together.This function implicitly converts the data values to a string and NULL values are converted to an empty string.

Type and execute the code below

SELECT CONCAT ('Liverpool', ' ' , 'is', ' ', 'the', ' ', 'best', ' football', ' ', 'team', ' ' ,'in', ' ' ,'the', ' ' ,'world')

SELECT Concat('Nikos',Null,' Kantzelis', ' Mr.')

SELECT CONCAT ('Usain', ' ' , 'Bolt', ' ' , 'came', ' ', 1, 'st', ' ' ,'in', ' ' , 'the', ' ' , 'Olympics')

All these strings in the first statement are now concatenated into one.

The same happens for the second statement but please note that the NULL value is converted to an empty string.

In the third statement the strings are concatenated.Please note  that CONCAT() function in this case will convert the value of 1 to its string equivalent.

Let’s try this new function in a table

Type and execute the code below

USE tempdb;
GO

CREATE TABLE [dbo].[Employee](
[ID] [int] IDENTITY(1,1) NOT NULL,
[FirstName] [nvarchar](30) NOT NULL,
[LastName] [nvarchar](30) NOT NULL,
[Country] [nvarchar] (30) NOT NULL,
[City] [nvarchar] (30) NOT NULL,
[PostalCode] [nvarchar] (30) NOT NULL,
[HireDate] [date] NOT NULL,
CONSTRAINT [PK_Employee] PRIMARY KEY CLUSTERED ([ID] ASC)
) ON [PRIMARY]

INSERT INTO Employee (FirstName, LastName,Country,City,PostalCode,HireDate)
VALUES
('Robbie', 'Fowler','USA', 'New York', '434343','1/10/2004'),
('Steven', 'Gerrard','Ireland','Dublin','75676','10/3/2000'),
('Michael', 'Owen', 'Britain','London','87677','2/4/2002')

SELECT CONCAT(FirstName, ' ' ,LastName) AS FULLNAME FROM Employee

I create a table and I insert some values in it. Then in the last statement I concatenate the first and the last name for all the rows in the table.

Hope it helps!!!

In this rather sort post I will demonstrate with a hands-on examples how to use 3 new conversion functions in SQL Server 2012, PARSE, TRY_CONVERT and TRY_PARSE.

You can have a look in another blog post of mine related to built-in SQL functions.You can have a look into the new logical functions introduced in SQL Server 2012, here .

You can use any edition of SQL Server of 2012 you want. SQL Server Express edition will do just fine.Download the Express edition, here .

In my case I will use Developer edition of 2012 for these examples.

I will look into the PARSE() function first.We can convert a string value to numeric or datetime format with the PARSE function.If that is not possible an error will be returned.We must say that this function relies on CLR to do the conversions.If there is no CLR an error will be returned.

This function takes a string value as the first argument and as a second argument there is the data type that we want to convert the value in the first argument.The third argument (optional) is the culture in which the string value will be formatted.

Type and execute the code below

SELECT PARSE ('THURSDAY, 30 JANUARY 2014' AS datetime2 USING 'en-us') AS NEWDAY
SELECT PARSE('€345,98' AS money USING 'fr-FR') AS Newmoney
SELECT PARSE('nick' AS datetime2 USING 'en-GB') AS Res

The first two statements will succeed but the third one will fail.

Let’s try this new function in a table

Type and execute the code below

USE tempdb;
GO

CREATE TABLE [dbo].[Employee](
[ID] [int] IDENTITY(1,1) NOT NULL,
[FirstName] [nvarchar](30) NOT NULL,
[LastName] [nvarchar](30) NOT NULL,
[Country] [nvarchar] (30) NOT NULL,
[City] [nvarchar] (30) NOT NULL,
[PostalCode] [nvarchar] (30) NOT NULL,
[HireDate] [date] NOT NULL,
CONSTRAINT [PK_Employee] PRIMARY KEY CLUSTERED ([ID] ASC)
) ON [PRIMARY]

INSERT INTO Employee (FirstName, LastName,Country,City,PostalCode,HireDate)
VALUES
('Robbie', 'Fowler','USA', 'New York', '434343','1/10/2004'),
('Steven', 'Gerrard','Ireland','Dublin','75676','10/3/2000'),
('Michael', 'Owen', 'Britain','London','87677','2/4/2002')

SELECT FirstName,LastName,PARSE(PostalCode as int) newpostalcode FROM Employee

I create a table and I insert some values in it. Then in the last statement I parse the postal code values from string to int.

The next function I will be looking into is TRY_PARSE.It is very similar with the PARSE function. We can convert a string value to numeric or datetime format with the TRY_PARSE function.If that is not possible an NULL value will be returned.We must say that this function relies on CLR to do the conversions.If there is no CLR an error will be returned.

This function takes a string value as the first argument and as a second argument there is the data type that we want to convert the value in the first argument.The third argument (optional) is the culture in which the string value will be formatted.

Type and execute the code below

SELECT TRY_PARSE ('THURSDAY, 30 JANUARY 2014' AS datetime2 USING 'en-us') AS NEWDAY
SELECT TRY_PARSE('€345,98' AS money USING 'fr-FR') AS Newmoney
SELECT TRY_PARSE('nick' AS datetime2 USING 'en-GB') AS Res

The first two statements will succeed but the third one will fail and we will receive a NULL value as a result.

Type and execute the code below

USE tempdb;
GO
SELECT FirstName,LastName,TRY_PARSE(PostalCode as int) newpostalcode FROM Employee
SELECT FirstName,LastName,TRY_PARSE(Country as int) newcountry FROM Employee

The first  statement will succeed but the second one will fail and we will receive NULL values as a result.

The final function I will be explaining is the TRY_CONVERT function.

TRY_CONVERT takes a value that is passed to it and tries to convert it to the specified data type.

If the cast succeeds, TRY_CONVERT returns the value as the specified data type.On the other hand if an error occurs, NULL is returned.if you request a conversion that is explicitly not permitted, then TRY_CONVERT will fail with an error.The first argument is the data type into which to cast the expression (value to be cast).

Type and execute the code below

SELECT TRY_CONVERT(datetime2, '12/12/2010') AS Result;
SELECT TRY_CONVERT(datetime2, '32/12/2010') AS Result;
SELECT TRY_CONVERT(int, 'nick' ) AS Res

The first statement will succeed but the second and third statement will fail and we will get as a result of that a NULL value.

Type and execute the code below

USE tempdb;
GO
SELECT FirstName,LastName,TRY_CONVERT(int,PostalCode) newpostalcode FROM Employee
SELECT FirstName,LastName,TRY_CONVERT(INT, Country) newpostalcode FROM Employee

The first  statement will succeed but the second one will fail and we will receive NULL values as a result.

Type and execute the code below

SELECT TRY_CONVERT(xml, 6677) AS Res;
GO

This statement will return an error.To recap we saw 3 new conversion functions introduced in SQL Server 2012.

Hope it helps!!!

In this rather sort post I will demonstrate with a hands-on examples how to use 2 new logical functions in SQL Server 2012, CHOOSE and IIF.

You can have a look in another blog post of mine related to built-in SQL functions.

You can use any edition of SQL Server of 2012 you want. SQL Server Express edition will do just fine.Download the Express edition, here

In my case I will use Developer edition of 2012 for these examples.

I will start with the CHOOSE function.This function returns an item from a specified set of  items for the specified index.You can think of it as an array kind of thing.

Launch SQL Server 2012 and in a new query window type and execute the code below

SELECT CHOOSE ( 3, 'Chelsea', 'Man Utd', 'Liverpool', 'Arsenal' ) AS 'The Best team';

But what happens if the index is not an integer? In that case the value is converted to an integer value.

Execute the code below

SELECT CHOOSE ( 3.3, 'Chelsea', 'Man Utd', 'Liverpool', 'Arsenal' ) AS 'The Best team';

You can use a variable to define the index value.

Execute the code below

declare @index as int = 3

SELECT CHOOSE ( @index, 'Chelsea', 'Man Utd', 'Liverpool', 'Arsenal' ) AS 'The Best team';

In this last example I will be using a table from the AdventureWorks2012 OLTP database.

You can download it from codeplex.

I will be using CHOOSE to concatenate values from the Person.Person table.Obviously this is not the best way to do this but I just want to demonstrate the CHOOSE function

Type and execute the code below

USE AdventureWorks2012
GO

SELECT CHOOSE(1,Title,FirstName,MiddleName,LastName) + ' ' + CHOOSE(2,Title,FirstName,MiddleName,LastName) + ' ' + CHOOSE(3,Title,FirstName,MiddleName,LastName) + ' .' + CHOOSE(4,Title,FirstName,MiddleName,LastName) FROM Person.Person;

Let’s move on to the IIF function.The IIF function returns one of two values depending if the first expression evaluates to TRUE or FALSE.If the first expression evaluates to TRUE the first value is displayed,if not the second value is displayed.

Type and execute the code below

SELECT IIF(DATEPART(WEEKDAY,GETDATE())=3 ,'Today is Tuesday','It is not Tuesday') as 'Day of the Week' ;

Please note that the start of the week is Sunday with index 1. So today as I am writing this post (Saturday night) the above script will return “It is not Tuesday” and that is because the first argument of the expression returned “FALSE”.

Ιn this last example I will use the Product table of the AdventureWorks database.In this example we want to produce a small report where every product that starts with ‘Touring’ should return back ‘Out Of Stock‘.

Type and execute the following code

SELECT [ProductID]
,[Name]
,[ProductNumber]
,[MakeFlag]
,[ListPrice]
,[Size]
,IIF((LEFT(Name,7)= 'Touring'), 'Out Of stock','In stock') as StockAvailability
FROM [AdventureWorks2012].[Production].[Product]

Please have a look at the results!!!

Hope it helps!!!!

I had no intention of writing such a post but I installed SQL Server 2012 Developer edition a few weeks back in my Windows 7 Ultimate SP1 laptop and it worked just fine.Then I thought it would be a good idea to write a post like that since a lot of people install the latest version of SQL Server.

Developer edition of SQL Server 2012 can be installed in a client operating system.

Enterprise edition can only be installed in server operating systems.You can download the Enterprise edition from this link.

If you are a MSDN subscriber (or you have purchased a valid license) you can still follow the steps below.

Before you download the Enterprise edition of SQL Server 2012 make sure you have read the Hardware and Software requirements

Being a MCT  I have a TechNet subscription and I have downloaded the Developer edition from my account.

Before I show you how to install the product I would like to talk a bit about the various editions of SQL Server 2012.

There are 3 main editions of SQL Server 2012

• Enterprise – Suitable for mission critical applications and data warehouses.All the features are included in this edition
• Business Intelligence – This is a brand new edition.It provides things like Rapid Data Discovery with Power View and a new feature called Data Quality Services.
• Standard – This is the basic edition and provides database,reporting and analytics capabilities.

If you want to check what features are supported by the various editions of SQL Server 2012 have a look at this msdn article.

There are also some special edition

• Web – This is basically the Standard edition.It is targeted towards serving data from websites,hosted websites.ISPs should definitely look to purchase licenses for this edition
• Developer – This has all the features of the Enterprise edition but it is only for demo purposes and development environment
• Express – This is a free edition and has only database engine features .There are some limitations of course.Each database has a maximum database size of 10 GB and the maximum memory utilised is 1 GB.

Obviously at some point the licensing issue will come up.Have a look here if you are interested.

Another important thing to note is that SQL Server 2012 is available in both 32 and 64 bit versions.

The SQL Server 2012 installation process is pretty similar with the one of SQL Server 2008.One notable difference is that BUILTIN\administrators and Local System (NT AUTHORITY\SYSTEM) are not automatically provisioned in the sysadmin fixed server role.

Let’s install SQL Server 2012, now.

Installation Steps

1) I run the SQL Server installer by starting SETUP.EXE

2) The SQL Server Installation Center will now display

3) Then I click on Installation

4) Then I click on New SQL Server stand-alone installation or add features to an existing installation

Have a look at the picture below

5)  The Setup Support Rules dialog displays.If there are any failures they must be corrected.In my case there none.Then I click OK

Have a look at the picture below

6) The next screen appears, the Product Updates screen. Then I click Next since there are no updates.

Have a look at the picture below

7) On the Setup Support Rules screen I have a look in the detailed report and I make a note of the warning.Problems that might occur in this stage must be corrected.

Have a look at the picture below

8) Then I click Next.On the Installation Type, Ι select Perform a new installation of SQL Server 2012. Notice that there is already a default instance of SQL Server installed.The default instance is SQL Server 2008 R2 Developer Edition. Τhere is also an instance of SQL Server 2008 R2 Express Edition and another instance of SQL Server 2012 Developer edition.You can have all these instances installed in one machine without any problem.

Have a look at the picture below

9) Then I click Next.On the Product Key screen, in my case there is already the product key.In your case you will have to type it in or select Specify a Free Edition (this will install SQL Server 2012 180 day evaluation copy)

Have a look at the picture below

10) Then I click Next. On the License Terms Page, Ι click I accept the license terms.

Have a look at the picture below

11) Τhen I Click Next.  On the Setup Role screen, click SQL Server Feature Installation and then I click Next. Have a look at the picture below

12) On the Feature Selection screen, I review the features I would like to install. When installing the Enterprise edition of SQL Server you see that there are many components that you could install.If you do not need to install SQL Server Analysis Services, SQL Server Integration Services,SQL Reporting Services then do not.Try (if this is an option for you) not to install SQL Server Reporting Services on the same server as the database engine. IIS and web services bring a lot of vulnerabilities to SQL Server.I have selected
· Database Engine Services
· SQL Server Replication
· Data Quality Services

Have a look at the picture below

13) Then I click the Next button.On the Installation Rules screen, click Next.

14) In the Instance Configuration screen I specify a named instance  “SQL2012Main” as the name.

Have a look at the picture below

15) Then I click Next. On the Disk Space Requirements screen, click Next.

16) In the Server Configuration screen Ι will configure the service accounts for the SQL Server instance. I can choose the security account the service will run and whether the service will start automatically or not.I will leave the defaults but you can choose whatever suits your needs.You can also set the collation for the server instance. In my case is Greek_CI-AS . You can change the collation of the server after the installation and you can set the collation at the database level.

Ηave a look at the picture below

17) Then I click Next. In the Database Engine Configuration screen, the authentication mode is set to Windows Authentication. Then I need to specify SQL Server Administrators Accounts and I click the Add Current User button.You can also have a look at the Data Directories tab.As we know it is a good practice to separate data files from log files but I will leave the default Data Directories.Have a look at the FILESTREAM tab.The Filestream option can be enabled at the point of installation, but can also be configured post installation. FILESTREAM enables SQL Server-based applications to store unstructured data, such as documents and images on the file system.

Have a look at the picture below

18) Then I click Next.In the Error Reporting screen, click Next.

19) In the Installation Configuration Rules screen, I click Next.In the Ready to Install screen, I did review the Installation Summary page.

Also I reviewed the configuration file path, which indicates the configuration.ini file used.

Have a look at the picture below

20) Then I click the Install button.Then installer will now install the SQL Server binaries.When the installation completes the Complete screen will appear.

Then I need to close the installer window and finally close the SQL Server Installation Center.

Installing SQL Server is a very straight-forward, smooth process and you will have no problem completing it.

Hope it helps!!!!

Now we are ready to go on with our hands-on example.I will be using the SQL Server 2012 Express Edition

If you have an instance of SQL Server 2012 running in your machine,simply connect to it.

In a new query window type

use master;
GO

CREATE SERVER ROLE MyCustomServerRole;

GRANT CONTROL SERVER TO MyCustomServerRole;

DENY ALTER ANY LOGIN TO MyCustomServerRole;
DENY ALTER ANY SERVER AUDIT TO MyCustomServerRole;
DENY ALTER ANY SERVER ROLE TO MyCustomServerRole;
DENY CREATE SERVER ROLE TO MyCustomServerRole;
DENY UNSAFE ASSEMBLY TO MyCustomServerRole;

CREATE LOGIN Mary WITH PASSWORD = '!@#$%^&';

EXECUTE AS LOGIN ='Mary';

CREATE LOGIN Jane WITH PASSWORD = '!@#$%^&abc';

SELECT * FROM sys.dm_exec_cached_plans;

CREATE SERVER ROLE JaneRole;

REVERT;

ALTER SERVER ROLE MyCustomServerRole ADD MEMBER Mary;

EXECUTE AS LOGIN ='Mary';

CREATE LOGIN Jane WITH PASSWORD = '!@#$%^&abc';

SELECT * FROM sys.dm_exec_cached_plans;

CREATE SERVER ROLE JaneRole;

Let me explain what I am doing with the T-SQL code above.

• In lines 4-6, I create a custom server role (MyCustomServerRole) and then grant it sysadmin permissions.
• In lines 8-12, I take some of these permissions away.
• In line 14 I create a SQL Server login,Mary.
• In line 15 I impersonate the login Mary and I execute lines 18-22 as Mary.
• All the statements in lines 18-22 fail since Mary has no server level permissions. That was expected.
• Then in line 24 I revert to being a sysadmin
• In line 26 I add Mary to the MyCustomServerRole and login again as Mary
• The statement in line 30 will fail (there is no such permission for that login) – Have a look at the DENY permission statements above
• The statement in line 32 will succeed
• The statement in line 34 will fail (there is no such permission for that login) – Have a look at the DENY permission statements above

Now le’ts see how to create a windows login. We need to create a windows user first. Right-click on Computer icon on the desktop and then select Manage, expand Local Users and Groups and then create New User.I create a windows user named “nikolaosk”. Put any password you want. Have a look at the picture below.

I am creating a new login from that windows user. Then I add the login to the user defined server role.

use master;
GO
CREATE LOGIN [FOFO-PC\nikolaosk] FROM WINDOWS;

ALTER SERVER ROLE MyCustomServerRole ADD MEMBER [FOFO-PC\nikolaosk];

Now I will launch another instance of SSMS and log in as “nikolaosk”

In a command prompt type runas  /user:nikolaosk “C:\Program Files (x86)\Microsoft SQL Server\110\Tools\Binn\ManagementStudio\Ssms.exe”

Then type the password.

In the new SSMS application (running as nikolaos), in a new query  window type again

use master;
GO
CREATE LOGIN Jane WITH PASSWORD = '!@#$%^&abc';

SELECT * FROM sys.dm_exec_cached_plans;

CREATE SERVER ROLE JaneRole;

SELECT * FROM sys.fn_my_permissions(NULL,'server')

The first statement will fail since there are no permissions for that operation for the MyCustomServerRole user defined role that login nikolaosk is member of.

The second statement will succeed but the third statement will fail since there are no permissions for that operation for the MyCustomServerRole user defined role that login nikolaosk is member of.If you want to see what kind of server level permissions login nikolaosk has, execute the last statement.

Close the second SSMS application. Now go back to the first SSMS application.

In a new query window type and execute the following statements.

use master;
GO

DROP LOGIN Mary;
DROP LOGIN [FOFO-PC\nikolaosk];

Disconnect from the server instance and then close SSMS.

Hope it helps!!!

But why we take backups in the first place? Disaster-Recovery is the first thing that comes to mind.We also use backups in Log Shipping,Database Mirroring,Replication. We do backups when we want to transfer our database from the development to the production server.Finally we take backups when moving or upgrading a database.

One could argue that we take backups when we want to manage the size of the transaction log.The only operation that clears the transaction log of a database that is in FULL or BULK logged recovery model is a transaction log backup. Some people are performing a backup before performing a database repair or an upgrade of any kind.

Before we proceed I would like to explain the various recovery models. In the Simple Recovery Model every time I have a checkpoint operation or the .ldf file is full, the .ldf file is automatically truncated.In this model the transaction log is not backed up.You can not have point in time data recovery with this model.Data is recoverable only to the most recent backup.

The Simple Recovery Model  is only suitable for small databases and databases with few transactions.We could use this recovery model if we want to have a small transaction log and for databases that are read-only or in development stage.

The Full Recovery Model is the default recovery model.In this model the transaction log grows rapidly since all database transactions are logged in at a very detailed level.If I have a catastrophe in my database I can rollback to the time where the catastrophe occurred.This recovery model is suitable for databases with thousands of transactions per day and where no loss of data is tolerated.

The Bulk Recovery Model works almost identically with the Full Recovery Model.It logs everything in extents=8pages and all bulk tasks (bcp,bulk insert) are logged minimally.We can go back to the time of failure but it will take more time to get the database back on-line.This model would be very suitable for databases with large bulk tasks.The disadvantage of this model is that you cannot go back at a specific point in time.

Having looked into the recovery models,let’s have a look on the various SQL Server backup types.

Let’s see what happens when we take a full database backup.The first thing to say is that you need always a full database backup.This is the base for all subsequent backups. You cannot have a differential backup,transaction log backup if you do not have at least one full backup.With a full database backup you backup all your data.You can have a full backup with all the recovery models that you choose to have for your database.Every full backup I perform in a database contains in it all the data previously fully backed up.
The backup operation is an online operation and the first thing it happens is to log that a backup was started. Then we write the data in the backup file.Then we do log any changes that occur as we backup.Finally the backing up of the data is finished and we log that fact.If you’re performing only full backups, you might lose some data in the event of a system crash—specifically, any changes made since the last full backup.

A differential backup is cumulative backup. It includes all data and structures that have changed since the last full backup, regardless of when that last full backup was made, or how many previous differential backups have been run.You can have a differential backup with all the recovery models.

The Transaction Log backup only works for full or bulk-logged recovery models.Each T-Log backup contains all the transactions that are in my database since the last T-Log backup.Always perform a full back up before you take a T-log backup.The time taken for a T-log backup is little and there is no performance penalty.An important thing to remember is that for every t-log back up I need the previous backup to have a restore.

I know a lot of people that are confused about the whole backup and restore process of a database. There are many issues that confuse people but the one I have found the most confusing for people is the use of Tail-log backup. Well, a tail-log backup is a transaction log backup that holds all the records that have not been backed up. So before we begin our restoration process,because we want to recover our SQL Server database to its latest point in time we must back up the tail of its transaction log.This will be the last back up.One thing that I want to immediately highlight is that you can not have a transaction log backup or tail-log backup unless the database is in either FULL or BULK-LOGGED recovery model.

How often you should back up a database depends on how long you have to restore it. In general, the more often you back up a database and the type of backup you take, the shorter the restore time.
You can tailor backups and restores for each database. The kind of backup you decide to use will depend on the size of the database and the amount of transactional activity.

Having only full backups is not a good strategy.If we have only one full database backup taken on 12 a.m every day for a database and our system crashes on 5 p.m, then we lose 5 hours of data.

If you can’t tolerate any data loss on restore, use the full plus log backup strategy.With this strategy we have a full database backup taken every day and then numerous T-Log backups at short periods of time. This method guarantees against data loss and works well for databases that are updated frequently. Although using this strategy increases your database’s complexity and maintenance, the total amount of time necessary to back up the database will decrease.So one might think that this is the appropriate restore strategy.At some point we will have thousands of files.It will take forever to restore thousands T-Log files.In order to have a good restore strategy for our average highly transactional database we need to include differential backups in our strategy. We could have for example 1 full backup every week, 7 differential backups every night and 336 (a week has 168 hours and I take 2 T-Log backups per hour) T-Log backups.

If you want to see a good backup (restore) strategy, have a look at this picture from sqlmag.com

Now we are ready to go on with our hands-on example.I will be using the SQL Server 2012 Express Edition

If you have an instance of SQL Server running in your machine,simply connect to it.

In an new query window type

USE master
GO

CREATE DATABASE sampledatabase
GO

ALTER DATABASE sampledatabase SET RECOVERY FULL

USE sampledatabase
GO

CREATE TABLE SampleData
(
id INT PRIMARY KEY ,
col1 tinyint NOT NULL ,
col2 BIT NOT NULL ,
col3 BIT NOT NULL ,
RegistrationDate DATETIME NOT NULL
)

GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94495, 25, 1, 1, '02-03-2008')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94496, 26, 1, 1, '02-04-2008')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94497, 27, 1, 1, '02-05-2008')

BACKUP DATABASE sampledatabase TO DISK ='C:\sqldata\backups\sampledata_full.bak' WITH INIT,STATS;
GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94498, 29, 1, 1, '02-03-2009')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94499, 28, 1, 1, '02-04-2009')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94500, 30, 1, 1, '02-05-2009')

SELECT * FROM sampledata

BACKUP DATABASE sampledatabase TO DISK = 'C:\sqldata\backups\sampledata_diff.bak' WITH DIFFERENTIAL
GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94501, 29, 1, 1, '02-03-2010')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94502, 28, 1, 1, '02-04-2010')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94503, 30, 1, 1, '02-05-2010')

BACKUP LOG sampledatabase TO DISK ='C:\sqldata\backups\sampledata_Log1.bak' WITH INIT,STATS;
GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94505, 29, 1, 1, '02-08-2010')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94506, 28, 1, 1, '02-09-2010')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94507, 30, 1, 1, '12-05-2010')

BACKUP LOG sampledatabase TO DISK ='C:\sqldata\backups\sampledata_Log2.bak' WITH INIT,STATS;
GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94508, 29, 1, 1, '02-28-2011')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94510, 28, 1, 1, '02-27-2011')

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (94511, 30, 1, 1, '12-12-2011')

BACKUP LOG sampledatabase TO DISK ='C:\sqldata\backups\sampledata_Log3.bak' WITH INIT,STATS;
GO

INSERT [dbo].[SampleData] ([id],[col1], [col2], [col3], [RegistrationDate])
VALUES (100000, 30, 1, 1, '12-02-2012')

SELECT TOP 5*
FROM sampledata
ORDER BY id DESC

USE master
GO

ALTER DATABASE sampledatabase SET OFFLINE WITH NO_WAIT;

--simulate the disaster by deleting the .mdf file

ALTER DATABASE sampledatabase SET ONLINE

--this will result in a nasty error since there is no database file

BACKUP LOG sampledatabase TO DISK ='C:\sqldata\backups\sampledata_TailLog.bak' WITH INIT,NO_TRUNCATE;
GO

RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_full.bak' WITH NORECOVERY, REPLACE
RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_diff.bak' WITH NORECOVERY

RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_Log1.bak' WITH NORECOVERY

RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_Log2.bak' WITH NORECOVERY

RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_Log3.bak' WITH NORECOVERY

RESTORE DATABASE sampledatabase FROM DISK='C:\sqldata\backups\sampledata_TailLog.bak' WITH NORECOVERY

RESTORE DATABASE sampledatabase WITH RECOVERY

USE sampledatabase
GO

SELECT TOP 5*
FROM sampledata
ORDER BY id DESC

Let me explain what I am doing in the T-SQL script above.

1) In lines 1-21 I create a sample database,set the recovery model to FULL and create a sample table in it.

2) In lines 23-30 I insert 3 records in the table.

3) In lines 32-33 I take a full database backup.

4) In lines 35-42 I insert 3 more records in the table.

5) In lines 46-47 I take a differential backup.

6) In lines 49-56 I insert 3 more records in the table.

7) In lines 58-59 I take another transaction log backup.

8) In lines 61-68 I insert 3 more records in the table.

9) In lines 70-71 I take another transactional log backup.

10) In lines 73-80 I insert 3 more records in the table.

11) In lines 82-83 I take another transaction log backup.

12) In lines 85-86 I insert a new record in the table.

13) In lines 88-90 Ι perform a simple “Select” to see the newly inserted record.

14) In lines 92-95 I take the database offline.Then I go to the local path and delete the .mdf file.In my case the path is C:\Program Files\Microsoft SQL Server\MSSQL11.SQLEXPRESS\MSSQL\DATA. As you can understand I am simulating a disaster.

15) In line 99 I try to take the database online but I receive a nasty error as expected.

16) In lines 103-104 I take a tail-log backup.This is the first step in the restore process.

17) In lines 106-117 I restore with the correct sequence (full,differential,3 t-log backups,final tail-log backup). Everything works fine.

18) In lines 122-124, I am looking for the last record (id=100.000) that was inserted after the last t-log backup but was picked up from the tail-log backup. So despite the catastrophe I had, my database is back online and there is no data loss.

Please note that not everyone has the permission to backup a database.This permission is limited by default to members of the sysadmin fixed server role,members of the db_owner and db_backupoperator fixed database role.

Hope it helps!!!

• If I have a database running on a (SQL Server 2008 R2 Developer Edition-Enterprise Edition) that has no enterprise features and take a backup of that database, I can restore it to a SQL Server 2008 R2 Standard edition istance.
• If I have a database that has enterprise features (data compression,transparent data encryption,partitioning,change data capture) and try to restore it to an SQL Server 2008 Standard Edition instance I will fail.

Let’s move on to our actual demonstration.In this example I will create a database and then create a table. I will insert some rows into the table. I will then compress the rows in the table.Data compression is an Enterprise feature only.Finally I will take a full database backup.

I will connect to this instance of SQL Server 2008 R2 Developer edition and in a new query window I will type

USE master;
GO

IF DATABASEPROPERTYEX ('MyDB', 'Version') > 0
DROP DATABASE MyDB;
GO

CREATE DATABASE MyDB
ON (NAME = MyDB_Data,
FILENAME = N'C:\SqlData\MyDB_data.mdf')
LOG ON (NAME = MyDB_Log,
FILENAME = N'C:\SqlData\MyDB_log.ldf');
GO

USE MyDB;
GO

CREATE TABLE MyCompressionTable (ID INT, phrase1 CHAR(100),phrase2 CHAR(100))
GO

INSERT INTO MyCompressionTable VALUES (1,
'SQL Server is the best RDBMS','I love T-SQL')

GO 15000

-- DATA_COMPRESSION = ROW
ALTER TABLE MyCompressionTable
REBUILD WITH (DATA_COMPRESSION = ROW);
GO

BACKUP DATABASE MyDB
TO DISK = N'C:\SqlData\MyDB.bak'
WITH INIT;
GO

I am going to connect to the SQL Server 2008 R2 Standard Edition running in my machine.I will connect to this instance and try to restore the backup I have just taken.

USE master;
GO

RESTORE DATABASE MyDB2
FROM DISK = N'C:\SqlData\MyDB.bak'
WITH MOVE N'MyDB_Data' TO N'C:\SqlData\MyDB_data2.mdf',
MOVE N'MyDB_Log' TO N'C:\SqlData\MyDB_log2.ldf',
REPLACE, STATS = 10;
GO

I will receive an error after executing the query above. You can see below (a part) the error message

RESTORE could not start database ‘MyDB2′.

RESTORE DATABASE is terminating abnormally.

Database ‘MyDB2′ cannot be started in this edition of SQL Server because part or all of object ‘compressed’ is enabled with data compression or vardecimal storage format. Data compression and vardecimal storage format are only supported on SQL Server Enterprise Edition.

Database ‘MYyDB2′ cannot be started because some of the database functionality is not available in the current edition of SQL Server.

To recap,if you’re not using the persistent Enterprise-only features (Partitioning, Data Compression, TDE,Change Data Capture), an Enterprise edition database can even be restored event to SQL Server Express edition, if of course is small enough.

I think DBAs should be very careful.It happens quite often (because of the cost) to buy licenses of Standard SQL Server.A DBA can have this instance as part of his disaster recovery strategy.When the disaster happens and then tries to restore the backup to the  SQL Server Standard Edition instance, that will  fail if the database has enterprise features.

There is a dynamic management view in SQL Server 2008 that can help us identify if a database has Enterprise features. This is the sys.dm_db_persisted_sku_features dynamic management view.

In any query window you can type the following query.

SELECT * FROM sys.dm_db_persisted_sku_features;

In my example the output was

Compression   100

Hope it helps!!!

You cannot take a backup of a SQL Server 2008R2 database and try to restore it to a SQL Server 2008 server.Similarly you cannot take a backup of a SQL Server 2012 database and restore it to a SQL Server 2008 R2 server.

You probably think that this is some sort of a bug.It is not.Each database has a physical number.SQL Server can understand a minimum and a maximum database physical version number.

I will create a short hands-on example to demonstrate that.I have an SQL Server express 2008 edition instance running on my machine.I also have SQL Server 2008 R2 Developer edition.

I will connect to this instance of SQL Server 2008 Developer edition and in a new query window I will type

USE master;
GO
IF DATABASEPROPERTYEX ('MyDB', 'Version') >; 0
DROP DATABASE MyDB;
GO

-- Create database
CREATE DATABASE MyDB
ON (NAME = MyDB_Data,
FILENAME = N'C:\SqlData\MyDB_data.mdf')
LOG ON (NAME = MyDB_Log,
FILENAME = N'C:\SqlData\MyDB_log.ldf');
GO

-- Take full database backup
BACKUP DATABASE MyDB
TO DISK = N'C:\SqlData\MyDB.bak'
WITH INIT;
GO

I have created a database and have taken a full database backup.

I am going to connect to the SQL Server 2008 Express edition running in my machine.I will connect to this instance and try to restore the backup.

USE master;
GO

RESTORE DATABASE MyDB2
FROM DISK = N'C:\SqlData\MyDB.bak'
WITH MOVE N'MyDB_Data' TO N'C:\SqlData\MyDB_data2.mdf',
MOVE N'MyDB_Log' TO N'C:\SqlData\MyDB_log2.ldf',
REPLACE, STATS = 10;
GO

I will receive an error after executing the query above

Have a look at the picture below

The only way I could solve my problem was to use another functionality through SSMS-Generate Scripts.I selected the MyDB database and then right-clicked on Tasks—>;Generate Scripts. The wizard pops up.I clicked Next on the first screen.In the next screen I selected “Script entire database and all database objects” .Then I clicked Next.I chose to save the script to a file and not to clipboard or query window. Then I clicked the Advanced button. I had to select a few options. Have a look at the picture below.

In your case you would probably need to select different options…Then I clicked OK and then Next.Then I clicked Next once more, the script was generated and then I closed the wizard by clicking Finish.Then I just had to run the new script in the SQL Server 2008 R2 developer edition.My database was finally” restored” in the production server.

Finally I want to mention that compatibility level has nothing to do with trying to restore a backup to an earlier version of SQL Server.If you set the compatibility level of a database to e.g compatibility level 90 or 100 that does not mean that an SQL Server 2012 database backup can be restored on a SQL Server 2008 database server.

Hope it helps!!!

Most DBAs  have to administer many SQL Server instances.It is not efficient to configure and manage them separately.DBAs try to enforce standards (e.g naming standards) on the instances they manage but it is very difficult to enforce them.

Policy-Based Management (PBM) is available in SQL Server 2008 Enterprise and Standard edition.

As it was mentioned earlier PBM allows you to define and enforce policies on SQL Server.I will show you with a hands-on example  how to achieve that.We will create a simple policy that will enforce a naming convention (All user defined stored procedures for all the databases in the server must begin with the prefix usp )

We need an instance of SQL Server running in our machine. I have SQL Server 2008 R2 Developer edition installed in my machine.

In case you do not have an instance of Enterprise edition or Developer edition  of SQL Server 2008 / R2, you can download it here . This is a trial version.

There are a number of terms that we need to define in order to begin to understand Policy-Based Management:

1. Target – an entity (a database, a table, an index) that is managed by PΒΜ
2. Facet – a predefined set of properties that can be managed
3. Condition – a property expression that evaluates to True or False, e.g the state of a Facet
4. Policy – a condition to be checked and/or enforced

Policy-Based Management is configured in SQL Server Management Studio (SSMS).

1) Launch SQL Server Management Studio and log in to the local instance of the SQL Server.

2) Navigate to the Object Explorer and expand the Management node and the Policy Management node.Υou will see the Policies, Conditions, and Facets nodes.

3)Right-click on Policies node and then select New Policy.Give your policy a name.

Have a look at the picture below

Then in the Check Condition dropdown click New Condition,fill in the name and the expression.Have a look in the picture below.

and then Click OK.

4) In the evaluation mode we select “On change:prevent” .This means that if some user makes a change that will cause a policy to evaluate to false, then this action will be prevented.

Have a look at the picture below

Finally we click OK on the policy window.

5)Now we need to test the new policy.In an new query window type

CREATE PROCEDURE test_SP1
 AS

PRINT 'this will violate a policy'

When you try to execute the procedure above (for any database in the server) you will receive an error message from SQL Server since there is a policy violation.

Have a look at the picture below

At this point I hope you have a good understanding of what PBM can do for enforcing policies in an SQL Server instance.

Hope this helps!!!!

The importance of automating tasks is profound.The following tasks could be automated

• Backing up the database
• Checking data integrity
• Dropping and re-creating indexes
• Transferring data

We can automate all these tasks to occur on a regular schedule.A common example is to automate the Backup of a database, every Saturday night e.g at 12.00 am.

For task automation we need SQL Server service (MSSQLSERVER) to be up and running. The Windows application log is the location of all operating system messages for the Windows operating system.MSSQLSERVER service is needed to write events to the Windows application log.  We can have a look into the Windows application log using the Windows Event Viewer.SQL Server Agent service must also be up and running.SQL Server Agent’s role is to take an action after a notification comes through the Windows application log.

Before I move on, let me talk a bit about SQL Server Agent. SQL Server Agent executes jobs and fire alerts. Some people are confused with the terms jobs and alerts.Jobs and alerts are defined separately and can be executed independently.A job can invoke an alert but the opposite is also possible.When we have alerts we can also have notifications of one or more operators by e-mail using Database Mail. Have a look in this post of mine to see how to configure Database Mail in SQL Server.

There are some steps involved when creating a job

• Create the job and its steps
• Create a schedule of the job execution if the job is not to be executed on demand
• Notify operators about the status of the job

We can define a job and its steps in various ways.We can use T-SQL statements in a job step. We can also use commands that start the execution of a utility like bcp command.In a job step we can also execute a program that is developed with Visual Basic or C#. We must always define what happens when a job step fails.A job step starts the next job step when it is successful.If a job step fails any job steps that follow will not be executed.We must define how many times a job step should be retried.

We can create a job using various ways

• We can launch SSMS and define a job through SSMS
• By using system stored procedures (sp_add,sp_add_jobstep)

Let’s begin with our hands-on examples

We need an instance of SQL Server running in our machine. I have SQL Server 2008 R2 Developer edition installed in my machine.

In case you do not have an instance of Enterprise edition or Developer edition  of SQL Server 2008 / R2, you can download it here . This is a trial version.

1) Launch SQL Server Management studio and connect to the desired instance of  SQL Server.

2) I will create a simple job that backs up the AdventureWorks database.You can follow my example by backing up any database you wish.

3) Expand SQL Server Agent,right-click Jobs and then click New Job.
The New Job window appears.I will fill in a name for the job and a description in the General tab

Have a look at the picture below

Please note that all jobs are enabled by default. SQL Server Agent disables jobs if the job schedule is defined  at a specific time that has passed.It also disables jobs that are defined on a recurring basis with an end date that has also passed.

4) Now click on the Steps tab and click New.The New Job Step dialog appears.We must enter a name for the job step.In the Type drop down list, I choose Transact-SQL script ( T-SQL ) because the backup of the database will be executed using T-SQL.In the Database drop down list I choose the master database.I do that because master system database must be current if we want to backup a database.

Have a look at the picture below

The T-SQL statement for backing up the database is

BACKUP DATABASE [AdventureWorks] TO
DISK = N'C:\Program Files\Microsoft SQL Server\MSSQL10_50.MSSQLSERVER\MSSQL\Backup\AdventureWorks.bak'
WITH NOFORMAT, NOINIT,  NAME = N'AdventureWorks-Full Database Backup', SKIP, NOREWIND, NOUNLOAD,  STATS = 10
GO

5) Now click on the Advanced tab to specify retry attempts and retry interval .Then click OK

Ηave a look at the picture below

6) Now we are ready to create a Job schedule.Each created job can be executed on demand or by using a schedule.A scheduled job can occur at a specific time or on a recurring schedule.Select the Schedules tab and click New.
I fill in a name for the schedule and the schedule type is recurring.It occurs every Monday each week at 9 am starting on 25/05/2012

Have a look at the picture below

Click OK on this window.Click OK once more to finish creating the job.

7) If you want to create a notification about a job status you need to create an operator first.I will use SSMS to create an operator.Expand SQL Server Agent node and ten right-click on Operators and then click New Operator.Enter the name of the operator and then specify one or more methods of notifying the operator.

Have a look at the picture below

8) To notify one or more operators after the job finishes we need to select the job we just created and select Properties and then we need to select the Notifications tab.Then we just need to check the corresponding objects.

Have a look at the picture below

Then click OK

9) If you want to find out information about a job you can have a look at the sysjobhistory table of the msdb database.You can also select a job and right-click on it and then click View History.The Log File Viewer dialog box shows the history log of the job.We can find information such as

• Whether the job completed successfully or not
• Duration of the job
• Operators who were notified
• Date and time when the job step occurred

10) Alerts as mentioned earlier can be used to respond to potential problems,system errors and user-defined errors.Information about execution of  jobs ans system error messages is stored in Windows application log.The SQL Server Agent reads the log and then compares the stored messages with the alerts defined for the system.If there is a match, SQL Server Agent fires the alert.Let’s create an alert that responds to a system error.Expand SQL Server Agent node, right-click Alerts and click New Alert.In the New Alert dialog box I enter the name (deadlock_alert) of the alert.I will create an alert that will respond to deadlocks.In the Error number I enter 1205
and select the alert type to be SQL Server event alert.

Have a look at the picture below

Now we need to define the response for the alert.Click on the Response page.I check the Notify operators and then in the operator list I select the operator I created earlier.

Have a look at the picture below

To recap SQL Server allows us to automate administrator tasks such as index maintenance and database backups.We can create jobs,operators,notifications and alerts. SQL Server Agent must be up and running to have automation.

Hope it helps!!!

Database Mail is a very useful feature.Both the Maintenance Plan Wizard and Maintenance Plan Designer have the ability to send e-mails to DBAs.

We can use the SQL Server Management Studio to set up Database Mail.

We need an instance of SQL Server running in our machine. I have SQL Server 2008 R2 Developer edition installed in my machine.

In case you do not have an instance of Enterprise edition or Developer edition  of SQL Server 2008 / R2, you can download it here . This is a trial version.

1) Launch SQL Server Management studio and connect to the desired instance of the SQL Server.

2) Navigate to the Management folder , right-click on Database Mail and select Configure Database Mail.
You will see the Splash screen and after you click Next , the wizard starts off with the Select Configuration Task

Ηave a look at the picture below

3) We select the option Set up Database Mail by performing the following tasks.Click on Next to continue

4)If Database Mail has not been enabled you will see a message like the one in the picture below.Click Yes

5) If you are interested on how to turn on Database Mail in SQL Server you can type the following T-SQL statement in a new query window

sp_CONFIGURE 'show advanced', 1
GO
RECONFIGURE
GO
sp_CONFIGURE 'Database Mail XPs', 1
GO
RECONFIGURE
GO

6) In the next screen you will be prompted to create a Database Mail Profile.

Have a look at the picture below

A profile is basically a collection of SMTP accounts that can be used by SQL Server to send messages.When SQL Server sends a message it is sent to the profile and then the profile is responsible for seeing that the e-mail has been delivered.

We must enter a profile name,an optional description and then we add and configure an SMTP account.

Ι add in the Profile name : SQL Server Admin

I add in the Description : This is a SQL Server Admin profile

Then click the Add button.The New Database Mail Account screen appears

Have a look at the picture below to see what one might enter in the available fields.

Ι use Basic Authentication which requires to provide a user name and a password. Then I click OK.You will return to the original mail profile screen.

7) We now need to continue with the Database Mail Configuration Wizard.Click on Next to reach the Manage Profile Security screen.

Now we must assign a profile as public as private.Only specific roles or users have access to private profiles.A public profile allows any user or role with access to send mail.I will keep things simple and I will create a public profile.Then click Next to move on to the Configure System Parameters screen

Have a look at the picture below

Then click Next on the wizard and then we move to the summary screen.We click Finish to close the Wizard.

Have a look at the picture below

Click Close.The profile and the SMTP account has been created

8)Now we need to check if Database Mail works correctly.We can send a test email.Select the Database Mail folder, right-click and select Send Test E-mail

Ηave a look at the picture below

In the To field  enter an email address and then click Send Test E-mail.It worked fine in my case and the e-mail was successfully sent.

You can use T-SQL to send an email. In a new query window type

USE msdb
GO
EXEC sp_send_dbmail @profile_name='SQL Server Admin',
@recipients='nkantzelis@q-training.gr',
@subject='Hello Admin!!!! Test message',
@body='This is a test message.
Database Mail works!!!!'

9) You can find information about sent emails,failed emails and more by typing the following T-SQL statements in a query window.

SELECT *
FROM sysmail_mailitems
GO

SELECT *
FROM sysmail_sentitems
GO

SELECT *
FROM sysmail_faileditems
GO

SELECT *
FROM sysmail_log
GO

Hope it helps!!!!!

I know there are samples of how to do that in BOL,MSDN and elsewhere but I thought that it would be a good idea to give it a try with hands-on examples.

Many people have found useful my style of posting. They feel they can follow the steps and master the technology/topic.

We all know that at some point our database is going to grow and possibly become huge. In a database that it is e.g 3 Terabytes in size, we could have tables with million of rows.

When we execute queries against tables with million of rows then we will probably get slow response times.We could partition the tables which means dividing the rows in various partitions based on column values.When you execute queries against tables that are partitioned may be much faster because SQL Server is able to search through a single partition to find the records needed instead of the entire table.

If you want to create a partitioned table you must.

1) Firstly, we have to select the table that must be partitioned.

2) Then we need to select the column on which the partition will take place.

3) Then we need to create a partition function. This is a function that specifies how to partition the table by defining partition key boundaries for each subset of data.

4) Then we need to create as partition scheme.It maps individual partitions to filegroups.

When creating a partition scheme we must bear in mind the following

• Each partition must map to a filegroup
• A single filegroup can contain more than one partition
• Each partition can have its own filegroup
• All partitions can map to the same filegroup
• Defining extra filegroups allows for adding new partitions in the future
• A partition scheme uses only one partition function

Let’s move on to the hands-on examples.

We need an instance of SQL Server running in our machine. I have SQL Server 2008 R2 Developer edition installed in my machine.

In case you do not have an instance of Enterprise edition or Developer edition  of SQL Server 2008 / R2, you can download it from here . This is a trial version.

We need a database as well. I will create a brand new database later on.

1) Launch SQL Server Management studio and connect to the desired instance of the SQL Server.We will create a new database and then we will create 4 filegroups that will map to 4 physical files.

In an new query window type the following T-SQL statements and execute them

USE master
GO
CREATE DATABASE UNIVERSITY
GO

ALTER DATABASE UNIVERSITY ADD FILEGROUP UNIVERSITY_DATA_1
GO
ALTER DATABASE UNIVERSITY ADD FILEGROUP UNIVERSITY_DATA_2
GO
ALTER DATABASE UNIVERSITY ADD FILEGROUP UNIVERSITY_DATA_3
GO
ALTER DATABASE UNIVERSITY ADD FILEGROUP UNIVERSITY_DATA_4
GO

ALTER DATABASE UNIVERSITY
ADD FILE
( NAME = UNIVERSITY1,
FILENAME = 'C:\DATA\UNIVERSITY_1.ndf',
SIZE = 100MB,
MAXSIZE = 300MB,
FILEGROWTH = 50MB)
TO FILEGROUP UNIVERSITY_DATA_1
GO

ALTER DATABASE UNIVERSITY
ADD FILE
( NAME = UNIVERSITY2,
FILENAME = 'C:\DATA\UNIVERSITY_2.ndf',
SIZE = 100MB,
MAXSIZE = 300MB,
FILEGROWTH = 50MB)
TO FILEGROUP UNIVERSITY_DATA_2
GO

ALTER DATABASE UNIVERSITY
ADD FILE
( NAME = UNIVERSITY3,
FILENAME = 'D:\DATA\UNIVERSITY_3.ndf',
SIZE = 100MB,
MAXSIZE = 300MB,
FILEGROWTH = 50MB)
TO FILEGROUP UNIVERSITY_DATA_3
GO

ALTER DATABASE UNIVERSITY
ADD FILE
( NAME = UNIVERSITY4,
FILENAME = 'd:\DATA\UNIVERSITY_4.ndf',
SIZE = 100MB,
MAXSIZE = 300MB,
FILEGROWTH = 50MB)
TO FILEGROUP UNIVERSITY_DATA_4
GO

2) I will create a table that has information regarding student registration.I will create a column in that table that is called “RegistrationDate”.This will be the partition column.Note that I have not created the table yet. This is a good column for the partition because we can partition on different date ranges e.g put the registration dates for the past year in one partition and the registration dates for the period 2009 in another.

3) Now we need to create the partition function.In the same or a new query window type the following T-SQL statements and execute them

USE University
GO

CREATE PARTITION FUNCTION pf_regdate (DATETIME)
AS RANGE RIGHT FOR VALUES ('2008-01-01','2009-01-01','2010-01-01')

--by running this command you can see in that partition will go a record with registration date of 2009-06-01

SELECT $PARTITION.pf_regdate('2009-06-01')

I create 4 partitions. All the values that are less than 01-01-2008 will go into their own partition (1st partition).All the values greater than 01-010-2010 will go into the last partition (4th partition)

4)  We now need to create the partition scheme.In the same or a new query window type the following T-SQL statement and execute it

CREATE PARTITION SCHEME ps_regdate AS PARTITION pf_regdate
TO (UNIVERSITY_DATA_1,UNIVERSITY_DATA_2,UNIVERSITY_DATA_3,UNIVERSITY_DATA_4)

I need one partition per filegroup at a minimum.

5) Now we must define that we want to create a table on a partition scheme for the specified column.In the same or a new query window type the following T-SQL statements and execute them

CREATE TABLE StudentRegistration
(
id INT IDENTITY(1, 1)NOT NULL,
discount tinyint NOT NULL ,
Paid BIT NOT NULL ,
Confirmed BIT NOT NULL ,
RegistrationDate DATETIME NOT NULL
CONSTRAINT pk_registration PRIMARY KEY (id,RegistrationDate)
) ON ps_regdate (RegistrationDate)

GO

ALTER TABLE StudentRegistration
WITH CHECK

ADD CONSTRAINT CK_discount
CHECK (discount >= 20 AND discount <= 50)

6 )Now let’s insert some data into the table.I will insert only 10 records but as I said in the beginning we apply partitioning on tables that have millions of rows.

In the same or a new query window type the following T-SQL statements and execute them

USE University

SET IDENTITY_INSERT StudentRegistration ON
INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed],

[RegistrationDate]) VALUES (94495, 25, 1, 1, '02-03-2008')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94496, 48, 1, 0, '02-03-2009')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94500, 42, 1, 1, '03-03-2005')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94501, 42, 1, 1, '04-03-2005')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94502, 32, 0, 1, '01-07-2010')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94502, 32, 0, 1, '01-07-2010')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94503, 32, 0, 1, '01-07-2011')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94504, 32, 0, 1, '01-07-2006')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94506, 32, 0, 1, '01-07-2008')

INSERT [dbo].[StudentRegistration] ([id], [discount], [Paid], [Confirmed], [RegistrationDate])
VALUES (94509, 32, 0, 1, '07-07-2009')

SET IDENTITY_INSERT StudentRegistration OFF

7) We can see the newly created partitions and the rows in each of them by executing the T-SQL statement below.

SELECT OBJECT_Name(object_id),
* FROM sys.partitions
WHERE OBJECT_Name(object_id) = 'StudentRegistration'

Have a look at the picture below to see what I mean

8) Νο Ι am going to demonstrate how to manage partitions.I am going to show you how to archive a partition into a separate table and then split an existing partition in two.

Let’s assume that we want to archive the records from the first partition (rows with registration date earlier than 1-1-2008) into a separate table.

Firstly I need to create the table and then move the records.Type the following T-SQL statements in a new query window and execute them

USE UNIVERSITY
GO

CREATE TABLE StudentRegistrationHistory
(
id INT IDENTITY(1, 1)NOT NULL,
discount tinyint NOT NULL ,
Paid BIT NOT NULL ,
Confirmed BIT NOT NULL ,
RegistrationDate DATETIME NOT NULL
CONSTRAINT pk_registration_history PRIMARY KEY (id,RegistrationDate)
) ON UNIVERSITY_DATA_1

ALTER TABLE dbo.StudentRegistration SWITCH PARTITION 1 TO StudentRegistrationHistory

--will see 3 records with registration date earlier than 1-1-2008
SELECT * FROM StudentRegistrationHistory

--all the other records remaining will be in the StudentRegistration table in the same partitions

SELECT * FROM StudentRegistration

I create the table-history/archive table.

I accomplish the move from partition 1 of the first table to StudentRegistrationHistory  by executing the statement

“ALTER TABLE dbo.StudentRegistration SWITCH PARTITION 1 TO StudentRegistrationHistory “

Then I execute SELECT statements to see that the move actually happened.Partition 1  exists but has no records

9) Moving on let’s demonstrate how to split an existing partition to two partitions. I want to change my last partition that holds records from 1-1-2010 onward to two partitions that have records from 1-1-2010 to 1-1-2011 and another one from 1-1-2011 to present.

Type the following T-SQL statements in a new query window and execute them

USE UNIVERSITY
GO

ALTER DATABASE UNIVERSITY ADD FILEGROUP UNIVERSITY_DATA_5
GO

ALTER DATABASE UNIVERSITY
ADD FILE
( NAME = UNIVERSITY5,
FILENAME = 'D:\DATA\UNIVERSITY_5.ndf',
SIZE = 100MB,
MAXSIZE = 300MB,
FILEGROWTH = 50MB)
TO FILEGROUP UNIVERSITY_DATA_5
GO

ALTER PARTITION SCHEME ps_regdate

NEXT USED UNIVERSITY_DATA_5

ALTER PARTITION FUNCTION pf_regdate()
SPLIT RANGE ('2011-01-01')

SELECT OBJECT_Name(object_id),
* FROM sys.partitions
WHERE OBJECT_Name(object_id) = 'StudentRegistration'

I create a new filegroup.Then I add a file to the newly created filegroup.Then I change the partition scheme  and finally split the records into the new partition using the SPLIT operator.

When I execute the last statement

(SELECT OBJECT_Name(object_id),
* FROM sys.partitions
WHERE OBJECT_Name(object_id) = ‘StudentRegistration’)

I see that I have 5 partitions now.Have a look at the picture below

10) Now I want to merge the records from the partition with values from  1-1-2008 to 1-1-2009 with the partition with values from 1-1-2009  to 1-1-2010.

Type the following T-SQL statements in a new query window and execute them

USE UNIVERSITY
GO

ALTER PARTITION FUNCTION pf_regdate()
MERGE RANGE ('2009-01-01')

SELECT OBJECT_Name(object_id),
* FROM sys.partitions
WHERE OBJECT_Name(object_id) = 'StudentRegistration'

I use the MERGE operator to achieve that by altering the partitioning function.When I execute the last statement I see that I have 4 partitions now.

Hope it helps!!!

In this post I will show you how collation affects the storage and behavior of your character data.Data is global nowadays.That means that at some point we will deal with databases that have data from different languages.That it why we must understand collation. Collation affects various factors.It affects the sort order of the data. It also affects the case and accent sensitivity for the various string comparisons.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition,  here.

During the installation of an SQL Server instance you are prompted to select the collation for the instance.

1) Launch SSMS and connect to an instance of SQL Server.Then select the instance of the SQL Server and select Properties.In the General tab you will see the collation of the server.Have a look at the picture below

2)We can specify the collation at a database level. Select the database and then select Properties and then Options from the popup window.

In that window we can specify the collation for the database .If not it will keep the same collation as the server.

Have a look at the picture below

Please note that we can specify the collation at the column level as well.

You cannot specify collation for variable and parameters.

3) In this first example I will show you how search results are affected in a table that uses the same collation (SQL_Latin1_General) but different case sensitivity and accent settings. I will use the TEMPDB database. In an new query window type and execute the following T-SQL code.

USE tempdb
GO

CREATE TABLE products
(
ID int IDENTITY (1,1),
Product_1 nvarchar(50) COLLATE SQL_Latin1_General_Cp1_CS_AS NULL,
Product_2 nvarchar(50) COLLATE SQL_Latin1_General_Cp1_CI_AS NULL,
Product_3 nvarchar(50) COLLATE SQL_Latin1_General_Cp1_CI_AI NULL
)

INSERT INTO products VALUES ('écrire au crayon','écrire au crayon','écrire au crayon'),
('baseball hat','baseball hat','baseball hat')

-- this will not bring back any results since the collation is case sensitive
SELECT * FROM products WHERE Product_1 ='BASEBALL HAT'

-- this will bring back  results since the collation is case insensitive
SELECT * FROM products WHERE Product_2='BASEBALL HAT'

-- this will bring back  results since the collation is case insensitive
SELECT * FROM products WHERE Product_3='BASEBALL HAT'

-- this will  bring back results since we use case sensitive search
SELECT * FROM products WHERE Product_1 ='baseball hat'

-- this will not bring back any results since the collation is accent sensitive
--note we have ecrire au crayon and not écrire au crayon
SELECT * FROM products WHERE Product_1 ='ecrire au crayon'

--this will fail as well since it is accent sensitive
SELECT * FROM products WHERE Product_2 ='ecrire au crayon'

--this will succeed since it is accent insensitive
SELECT * FROM products WHERE Product_3 ='ecrire au crayon'

--this will succeed since it is accent sensitive
SELECT * FROM products WHERE Product_1 ='écrire au crayon'

In this t-sql snippet we create a table that has 3 columns with same collation (SQL_Latin1_General) but different case sensitivity and accent settings.Then we insert some values. Then I perform some SELECT operations and in some cases I am successful and I get back results but in some other cases I am not. Please have a look at the comments in the code.

4) In this example I will show you why we get errors when we join tables with different collations. I will use the TEMPDB database. In an new query window type and execute the following T-SQL code.

USE tempdb
GO

CREATE TABLE footballers(
[id] [int] IDENTITY(1,1) NOT NULL,
[firstname] [varchar](50) NOT NULL,
[lastname] [varchar](50) NOT NULL,
[isActive] [bit] NULL,
teamcode[varchar](50) NOT NULL
)

INSERT [dbo].[footballers] ( [firstname], [lastname], [isActive],teamcode) VALUES ( N'Robbie', N'Fowler', 1,'Liv')
INSERT [dbo].[footballers] ( [firstname], [lastname], [isActive],teamcode) VALUES ( N'Steven', N'Gerrard', 1,'Liv')
INSERT [dbo].[footballers] ( [firstname], [lastname], [isActive],teamcode) VALUES ( N'Kenny', N'Dalglish', 0,'Liv')
INSERT [dbo].[footballers] ( [firstname], [lastname], [isActive],teamcode) VALUES ( N'John', N'Terry', 0,'Che')

CREATE TABLE Team(

code [varchar](50) NOT NULL,
name [varchar](50)  NOT NULL
)

INSERT Team(code,name) VALUES ('Liv','Liverpool')
INSERT Team(code,name) VALUES ('Che','Chelsea')

SELECT * FROM footballers JOIN Team
ON footballers.teamcode = team.code

CREATE TABLE GreekTeam
(
code [nvarchar](50) COLLATE Greek_100_BIN
NOT NULL ,
name [nvarchar](50) COLLATE Greek_100_BIN
NOT NULL
)

INSERT GreekTeam(code,name) VALUES (N'ΠΑΟ',N'Παναθηναικός')
INSERT GreekTeam(code,name) VALUES (N'ΟΣΦΠ',N'Ολυμπιακός')

SELECT * FROM GreekTeam

--this will fail

SELECT * FROM footballers JOIN GreekTeam
ON footballers.teamcode = GreekTeam.code

I create a table Footballers and I insert some data in it. I create a table called Team and I insert some data in it. Then I execute a JOIN statement and I get back results.

Then I create another table GreekTeam that uses the Greek_100_BIN collation.I insert some values in it.Then I try to execute another JOIN statement between Footballers table and the GreekTeam table. This results in an error, “Cannot resolve the collation conflict between “Greek_100_BIN” and “Latin1_General_CI_AI” in the equal to operation.”

This happens because the two tables do not use the same code page and it can’t compare the two values because they are stored differently.

5) Let’s have a look on another example and find out how collation affects the sorting of data.

In a new query window type the following T-SQL statements and execute them

USE TempDB
CREATE TABLE MyTable
(col1 varchar(10) COLLATE Latin1_General_CS_AS NULL,
col2 varchar(10) COLLATE Latin1_General_BIN)
GO

USE TempDB
INSERT MyTable
VALUES ('Alpha','Alpha')
INSERT MyTable
VALUES ('Bravo','Bravo')
INSERT MyTable
VALUES ('Charlie','Charlie')
INSERT MyTable
VALUES ('alpha','alpha')
INSERT MyTable
VALUES ('bravo','bravo')
INSERT MyTable
VALUES ('charlie','charlie')
GO

SELECT col1
FROM MyTable
ORDER BY col1 ASC

SELECT col2
FROM MyTable
ORDER BY col2 ASC

We create a table with 2 different columns and different collations.Then we insert some values. Finally we execute 2 select statements.When you execute them you will see that the data is returned back to us from the relational engine of SQL Server with different sort order.

Hope it helps!!!

This is not a post that experienced/advnanced DBAS/developers should read. The target audience is beginners (Level 100).

Most beginners in SQL Server  do not understand the concept of filegroups and why we need them.

Ι am going to start with a short introduction on data files,log files.

We all know that our data is stored eventually in operating system files, the same way the data we write in our Word documents are stored in .docx files in the physical disk.Every SQL Server database has two operating system files types:

• a data file
• a log file

A data file contains data and objects such as tables,indexes,stored procedures and views. Every database has one primary data file.The recommended file name extension for primary data file is .mdf.We can have secondary data files.The recommended file name extension for secondary data files is .ndf

The transaction log files hold the log information that is needed to recover the database. There must be at least one log file for each database. The recommended file name extension for transaction logs is .ldf.

Filegroups are named collections of files and are used to increase database performance,simplify data placement and administrative tasks such as backup and restore operations.

Filegroups enable data placement, because we can create a table in a specific filegroup. This improves performance, because all I/O for a specific table can be directed at a specific disk.

If you have tables that are frequently used together, you should put them on separate filegroups and physical drives.A good example could be  to put tables in the same join queries in different filegroups.

If you have a heavily used table, it can be put on one file in one filegroup, located on one disk, and the other less heavily accessed tables in the database can be put on the other files in another filegroup, located on a second disk.You can also put  the non-clustered indexes of a table in one filegroup located on one disk and the table in another filegroup on another disk.

Files cannot be moved to different filegroups and a file cannot be a member of more than one filegroup.We cannot have log files in filegroups.Filegroups can contain only data files.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition,  here.

Launch SSMS and connect to an instance of SQL Server. In a new query window type the following T-SQL statements and execute them

USE master
GO
CREATE DATABASE TEST
GO

ALTER DATABASE TEST ADD FILEGROUP TEST_DATA_1
GO
ALTER DATABASE TEST ADD FILEGROUP TEST_DATA_2
GO

ALTER DATABASE TEST
ADD FILE
( NAME = TEST1,
FILENAME = 'C:\DATA\TEST_1.ndf',
SIZE = 1MB,
MAXSIZE = 10MB,
FILEGROWTH = 1MB)
TO FILEGROUP TEST_DATA_1
GO

ALTER DATABASE TEST
ADD FILE
( NAME = TEST2,
FILENAME = 'd:\data\TEST_2.ndf',
SIZE = 1MB,
MAXSIZE = 10MB,
FILEGROWTH = 1MB)
TO FILEGROUP Test_DATA_2
GO

USE TEST
GO

CREATE TABLE TABLE1
(
ID INT IDENTITY(1,1),
NAME VARCHAR(100),
CONSTRAINT PK_TAB1 PRIMARY KEY(ID)
) ON TEST_DATA_1

GO

CREATE TABLE TABLE2
(
ID INT IDENTITY(1,1),
NAME VARCHAR(100),
CONSTRAINT PK_TAB2 PRIMARY KEY(ID)
) ON TEST_DATA_2

GO

First we create a database that is called TEST.Then we create two filegroups (Test_DATA_1,Test_DATA_2).Then we add two secondary data files to each of the filegroups. Then we have this scenario. We want to create two tables that we know that will be heavily accessed by our application.We want to place them in different filegroups.The last two SQL batches show how to accomplish that.

When I select the TEST database, right-click on it,select Properties and the Files and then Filegroups I have a visual representation of the operations I just issued.

Have a look below to see what I mean

Hope it helps!!!

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition,  here.

Launch SSMS and connect to an instance of SQL Server. In a new query window type the following T-SQL statements and execute them

USE tempdb
GO

CREATE TABLE tblUnsparsed
(
ID INT IDENTITY(1, 1) ,
OrderID  INT ,
Size VARCHAR(100) ,
OrderDate DATETIME
)
GO

CREATE TABLE tblSparsed
(
ID INT IDENTITY(1, 1) ,
OrderID  INT SPARSE ,
Size VARCHAR(100) SPARSE ,
OrderDate DATETIME SPARSE
)
GO

INSERT INTO tblUnsparsed VALUES (NULL,NULL, NULL)
GO 200000

INSERT INTO tblSparsed VALUES (NULL, NULL, NULL)
GO 200000

SELECT * FROM tblUnsparsed

SELECT * FROM tblSparsed

sp_spaceused 'tblUnsparsed'
GO
sp_spaceused 'tblSparsed'
GO

DROP TABLE tblSparsed
GO
DROP TABLE tblUnsparsed
GO

I create two tables.The second table has the SPARSE statement in the column definition.I insert 200.000 rows of NULL values in both of the tables.Then I check to see if the rows were inserted.200.000 rows were inserted in both tables.Then I use the sp_spaceused system stored procedure to find out how much space was consumed.

You will see a big difference when SQL Server returns the results.

Have a look in the picture below to see what I got

We cannot apply the SPARSE statement to columns where their data types are :image,text,geography,geometry.We cannot apply rules or defaults to SPARSE columns. We cannot have Data Compression and Merge replication does not work either.In case the SPARSE columns have data in them they will take 4 more bytes than the normal column. So please be careful when you use SPARSE columns.

Hope it helps!!!!

I will show you the various ways I know on how to accomplish that.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition,  here.

• This is the obvious way (most people should know that even if they are beginners in SQL Server).

I choose the database, in this case AdventureWorks2008R2, select Properties, then select Options.In this case I see that the recovery model is FULL

Have a look at the picture below

• There is a way to find out the recovery model for all databases from SSMS. Select the Databases node.From the menu select View and then Object Explorer Details. Have a look in the picture below to see the output.You must select from the available columns to show the Recovery Model column
  

• We can find out the Recovery Model for each database in the SQL Server using. In a query window type and then execute the following T-SQL code

USE master
GO

SELECT name AS [DBName],
recovery_model_desc AS [Recovery_Model]
FROM sys.databases
GO

• We can find out the recovery model for a specific database by typing and executing the following T-SQL statements in a new query window

USE master
GO

SELECT 'AdventureWorks2008R2' AS [DBName],
DATABASEPROPERTYEX('AdventureWorks2008R2', 'Recovery')
AS [Recovery_Model]
GO

Hope it helps!!!

Have a look in this post of mine to find out more information about transactions.A lot of people are confused on how a Rollback statement impacts a nested transaction.

I will move on with my hands-on examples.I will use the tempdb database.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition from here.

Launch SSMS and connect to an instance of SQL Server.In a new query window type

USE TempDB
GO
CREATE TABLE TranTable (ID INT, name VARCHAR(100))
GO

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID,  name)
VALUES (1, 'Nick');

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID, name)
VALUES (2, 'George');

SELECT ID, name
FROM TranTable;

--we issue a rollback

ROLLBACK

--we have no values in the table

SELECT ID, name
FROM TranTable;

--this is 0
--nested transaction rollsback to the outmost transaction

SELECT @@TRANCOUNT tranCount;

DROP TABLE TranTable
GO

I am creating a table and I insert some values in the table by starting explicitly a transaction. At the same time I query the @@Trancount global variable.It basically returns the number of  BEGIN TRANSACTION statements that have occurred on the current connection.Then I do the same in a second explicit transaction (lines 17-20).@@Trancount is 2 at this point.Then we issue a Rollbackstatement. Then  we issue a “SELECT ID, name FROM TranTable;”. We notice that there is no data in the table and the @Trancount is 0 at this point.This means that a Rollback statement rollsback to the outmost transaction that was defined. I want to show a similar example where I use the exact same T-SQL as before but I commit the transaction before I issue a rollback statement .In a new query window type

USE TempDB
GO
CREATE TABLE TranTable (ID INT, name VARCHAR(100))
GO

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID,  name)
VALUES (1, 'Nick');

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID, name)
VALUES (2, 'George');

COMMIT TRANSACTION

SELECT ID, name
FROM TranTable;

--we issue a rollback

ROLLBACK

--we have no values in the table

SELECT ID, name
FROM TranTable;

--this is 0
--nested transaction rollsback to the outmost transaction

SELECT @@TRANCOUNT tranCount;

DROP TABLE TranTable
GO

Execute lines 1-20.Then execute lines 22-25. We commit the transaction and then we check to see if the 2 rows reside on the table. @@Trancount is 2 at this point.Then we issue a rollback statement.Then we issue a “SELECT ID, name FROM TranTable;”. We notice that there is no data in the table and the @Trancount is 0 at this point.This means that a Rollback statement rollsback again to the outmost transaction that was defined.

In this last example I will demonstrate how we can avoid rolling back to the outmost transaction by using savepoints.

In a new query window type

USE TempDB
GO
CREATE TABLE TranTable (ID INT, name VARCHAR(100))
GO

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION Tran1
SAVE TRANSACTION Tran1
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID,  name)
VALUES (1, 'Nick');

SELECT ID, name
FROM TranTable;

BEGIN TRANSACTION Tran2
SAVE TRANSACTION Tran2
SELECT @@TRANCOUNT tranCount;
INSERT INTO TranTable (ID, name)
VALUES (2, 'George');

SELECT ID, name
FROM TranTable;

--we issue a rollback to a savepoint

ROLLBACK TRANSACTION Tran2

--we have one row in the table

SELECT ID, name
FROM TranTable;

--@@TRANCOUNT  is 2

SELECT @@TRANCOUNT tranCount;

ROLLBACK TRANSACTION Tran1

SELECT ID, name
FROM TranTable;

--@@TRANCOUNT  is 2

SELECT @@TRANCOUNT tranCount;

ROLLBACK

--@@TRANCOUNT  is 0

SELECT @@TRANCOUNT tranCount;

DROP TABLE TranTable
GO

Execute lines 1-22. We issue the same statements as before but as you can clearly see we save the first and second explicit transactions (Tran 1, Tran2).We then confirm that we have 2 rows in the table.Then we issue a rollback statement to the specific savepoint in line 30.Then we see that we have the first row still in the table by executing the lines 34-35.The @@TRANCOUNT is 2 at this point. Then we issue a rollback statement to a specific savepoint in line 42.Then we see that there are no rows in the table by executing the lines 42-43.The @@TRANCOUNT is 2 at this point.Then we issue a ROLLBACK statement. The @@TRANCOUNT is 0 at this point.This is the only way (savepoints) to return to a point within our transaction when issuing Rollback statements to those savepoints  .

Hope it helps!!!

We use DENY and GRANT  T-Sql commands to give permissions on a securable (e.g table) to a principal (a custom user we have created).Most developers are familiar with their use but there are some special cases I would like to point out.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition from here.

Launch SQL Server and in a new query window type the following

USE master
GO

CREATE DATABASE TestDB
GO

USE TestDB
GO

CREATE USER John WITHOUT LOGIN;
GO

CREATE TABLE Person (ID int, name VARCHAR(50), lastname VARCHAR(50))
GO

INSERT INTO Person VALUES (1,'george','Harisson'), (2,'Jill','Diamond'), (3,'Jane','Littlewood')
GO

SELECT * FROM Person

GRANT SELECT ON Person to PUBLIC

DENY SELECT (lastname) ON Person TO john
GO

EXECUTE AS user='John'

--this will fail
SELECT * FROM Person
GO
--this will work
SELECT ID, name FROM Person
GO

REVERT

GO

REVOKE SELECT (lastname) ON Person TO john

We create a sample database, a new user John  and a new table, Person.Then we insert some values to the Person table.Then we deny (DENY) the SELECT operation for the lastname column of the Person table to user John.Then we change the user to John. Then we do a Select * from Person , which fails but the Select ID,name from Person operation succeeds for the user John.User John has no DENY permissions for the ID and name columns so the SELECT operation succeeds.Then we do a REVERT operation and REVOKE the SELECT lastname permission on table Person to user John.

Let’s have a look on a second example

In the same query window type

DENY SELECT ON Person TO john

GRANT SELECT (name) ON Person TO john
GO

EXECUTE AS user='john'
--this will fail
SELECT * FROM Person
--this will succeed
SELECT name FROM Person
REVERT

USE master
GO
DROP DATABASE TestDB

We issue a DENY statement for a SELECT operation on the Person table for the John user.Then we issue a GRANT statement for a SELECT operation on the name column of the Person table.Then we change the user to John.Then we do a Select * from Person, which fails but the Select name from Person operation succeeds for the user John.So in this case we see that a GRANT statement takes precedence over DENY when it comes to column-level GRANT permissions.

Finally we issue a REVERT operation to revert to the dbo user and then we drop the TestDB database.

Hope it helps!!!

Some people believe that having duplicate indexes will improve the performance. There will be no performance gain for SELECT statements and there will be a performance degradation for INSERT,UPDATE,DELETE statements.

Have a look in this post if you want to learn about the possible dangers of over indexing your tables.

We will need a database to work with. I will use AdventureWorks2008R2. You can download it here .

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition from here.

I will create a new empty table in the AdventureWorks2008R2 .I will call it NewPerson and it will be based on the Person table and I will insert the same data that resides in the Person table to the NewPerson tables.There will be no indexes on the NewPerson table.I will use STATISTICS TIME ON,STATISTICS IO ON to measure the time needed to insert the rows to the table together with the IO operations.Then I will truncate the table. I will create multiple duplicate indexes on the FirstName column and I will insert the same data again in the table. I will then compare the time needed to insert rows in the table without any indexes and a table with multiple duplicate indexes.

Launch SQL Server Management Studio and log in the instance of SQL Server that AdventureWorks2008R2 is attached. In an new query window type

USE [AdventureWorks2008R2]
GO

SET NOCOUNT ON

SELECT *
INTO [Person].[NewPerson]
FROM [Person].[Person]
WHERE 3 = 4
GO

SET STATISTICS TIME ON
SET STATISTICS IO ON
GO

INSERT INTO [AdventureWorks2008R2].[Person].[NewPerson]
([BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate])
SELECT    [BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate]
FROM [Person].[Person]
GO

SELECT * FROM Person.NewPerson

Have a look at the statistics output. In my case I got the results shown in the picture below

Now I will truncate the table and I will create non-clustered duplicate indexes on the NewPerson table for the FirstName column.Then I will insert the same data again and make a note of the statistics again.Type the following in the same query window.

TRUNCATE TABLE [Person].[NewPerson]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName1]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName2]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName3]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName4]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName5]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

INSERT INTO [AdventureWorks2008R2].[Person].[NewPerson]
([BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate])
SELECT    [BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate]
FROM [Person].[Person]
GO

Have a look again at the statistics output. In my case I got the results shown in the picture below.
They are significantly greater (140% or higher).

The NewPerson table has 19972 rows only. If it had 2-3 million rows the elapsed time in the second case it would be 20(2000%) times greater.In any case drop duplicate indexes.

Hope it helps!!!!

Some people believe that this is not possible and think that if you want to enforce a unique index on a column, that column must have only one NULL value.I will show you that with the help of filtered indexes we can have unique indexes on columns with multiple NULL values.

Have a look in this post if you want to learn about the possible dangers of over indexing your tables.

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition from here.

In this example I will use tempdb.Launch SQL Server Management Studio and log in  SQL Server. In an new query window type

USE tempdb
 GO

CREATE TABLE MyTable
 (
 id INT IDENTITY(1, 1)
 PRIMARY KEY ,
 team VARCHAR(15)
 )

INSERT  MyTable
 VALUES  ( 'Liverpool' ),
 ( 'Barca' ),
 ( 'Real' ),
 ( 'Man Utd' )

SELECT * FROM MyTable

CREATE UNIQUE INDEX UNQ_team_MyTable ON MyTable(team)
 GO

INSERT MyTable Values (NULL)
 GO

Select * from MyTable

--this will create an error
 INSERT MyTable Values (NULL)
 GO

I create a table and I insert some values. Then I create a unique index on a column and insert a NULL value in it. So far so good. If I insert a second NULL value in the same table I get an error.In the same query window type

DROP INDEX  MyTable.UNQ_team_MyTable
 GO

--filtered index
 CREATE UNIQUE INDEX UNQ_team_MyTable ON MyTable(team)
 WHERE team IS NOT NULL
 GO
 --this will not fail
 INSERT mytable Values (NULL)
 GO
 --this will not fail
 INSERT mytable Values (NULL)
 GO
 --this will not fail
 INSERT mytable Values (NULL)
 GO
 --this will not fail
 INSERT mytable Values (NULL)
 GO

Select * from mytable
 GO

--this will fail
 INSERT mytable
 VALUES ('Barca')

--this will succeed

INSERT mytable Values (NULL)
 GO

I drop the index and create a filtered index on the column where the value is different than NULL.Then I insert multiple NULL values in the team column with no problem at all. If I try to insert a value in the team column of the table that already exists I will get an error from SQL Server. Note that filtered indexes are introduced in SQL Server 2008.Filters can only be applied to non-clustered indexes.

Hope it helps!!!!

I will beginning a series of small posts looking into the various aspects of indexes hoping to shed some light on their use and misuse.A lot of times I find out that every time a DBA or developer detects a performance degradation tends to create indexes on the tables of the backend (SQL Server in this case).If there is a problem with performance it does not mean that index creation is always the answer. Many reasons could cause performance bottlenecks (e.g the hardware cannot withstand the workload,excessive blocking or little reuse of the execution plans)

In this post I will show you why it is a bad idea to over index your SQL Server tables.Over indexing consumes unnecessary disk space and the query optimiser may use the less efficient index thus the less efficient execution plan.

We will need a database to work with. I will use AdventureWorks2008R2. You can download it here .

We will need an instance of SQL Server running in our machine.You can download and install the free SQL Server Express edition from here.

I will create a new empty table in the AdventureWorks2008R2 .I will call it NewPerson and it will be based on the Person table and I will insert the same data that resides in the Person table to the NewPerson tables.There will be no indexes on the NewPerson table.I will use STATISTICS TIME ON to measure the time needed to insert the rows to the

Launch SQL Server Management Studio and log in the instance of SQL Server that AdventureWorks2008R2 is attached. In an new query window type

USE [AdventureWorks2008R2]
GO

SET NOCOUNT ON

SELECT *
INTO [Person].[NewPerson]
FROM [Person].[Person]
WHERE 3 = 4
GO

SET STATISTICS TIME ON
GO

INSERT INTO [AdventureWorks2008R2].[Person].[NewPerson]
([BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate])
SELECT    [BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate]
FROM [Person].[Person]
GO

SELECT * FROM Person.NewPerson

Have a look at the statistics output. In my case I got the results shown in the picture below

Now I will truncate the table and I will create a clustered and non clustered indexes on almost every column of the NewPerson table.Then I will insert the same data again and make a note of the statistics again.Type the following in the same query window.

TRUNCATE TABLE [Person].[NewPerson]
GO

ALTER TABLE [Person].[NewPerson]
ADD CONSTRAINT [PK_BusinessEntityID]
PRIMARY KEY CLUSTERED
([BusinessEntityID] ASC,
[LastName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_NewPersonType]
ON [Person].[NewPerson]
([PersonType] ASC) ON [PRIMARY]
GO
CREATE NONCLUSTERED INDEX [IX_NameStyle]
ON [Person].[NewPerson]
([NameStyle] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_FirstName]
ON [Person].[NewPerson]
([FirstName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_MiddleName]
ON [Person].[NewPerson]
([MiddleName] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_NSuffix]
ON [Person].[NewPerson]
([Suffix] ASC) ON [PRIMARY]
GO
CREATE NONCLUSTERED INDEX [IX_EmailPromotion]
ON [Person].[NewPerson]
([EmailPromotion] ASC) ON [PRIMARY]
GO

CREATE NONCLUSTERED INDEX [IX_rowguid]
ON [Person].[NewPerson]
([rowguid] ASC) ON [PRIMARY]
GO
CREATE NONCLUSTERED INDEX [IX_ModifiedDate]
ON [Person].[NewPerson]
([ModifiedDate] ASC) ON [PRIMARY]
GO

INSERT INTO [AdventureWorks2008R2].[Person].[NewPerson]
([BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate])
SELECT    [BusinessEntityID]
,[PersonType]
,[NameStyle]
,[Title]
,[FirstName]
,[MiddleName]
,[LastName]
,[Suffix]
,[EmailPromotion]
,[AdditionalContactInfo]
,[Demographics]
,[rowguid]
,[ModifiedDate]
FROM [Person].[Person]
GO

Have a look again at the statistics output. In my case I got the results shown in the picture below.
They are significantly greater (40% or higher).

In this case we see that over indexing does not help the performance of the system.We should drop the ones we do not need.   The NewPerson table has 19972 rows only. If it had 2-3 million rows the elapsed time in the second case it would be 10(1000%) times greater.

Hope it helps!!!!