Recovery Models in SQL Server

Here we are going to discuss about database recovery model in SQL Server.

Recovery Models:
1. Simple
2. Bulk Logged
3. Full

Simple Recovery Model:
a)All the operations are not logged in Log file.
b)Log file size won't grow much.
c)Suitable for LOCAL and STAGE Environment.
d)Can't recover the database to certain point in time state.

Bulk Logged Recovery Model:
a)Only minimal information logged in log file during BULK operation. So, its also called as "Minimal Logging"
b)Log File won't grow much during bulk operation.
c)Suitable during bulk operation, migration.
d)Can't recover the bulk operation data.

Full Recovery Model:
a)All the operation logged in log file.
b)Log file size grow rapidly depends upon transactions.
c)Suitable for PRODUCTION environment.
c)Can recover the database to certain point in time (if we followed proper backup plan).

Here is script to switch to different recovery model.

 ALTER DATABASE testdb SET RECOVERY SIMPLE  
 GO  
 ALTER DATABASE testdb SET RECOVERY BULK_LOGGED  
 GO  
 ALTER DATABASE testdb SET RECOVERY FULL  
 GO  

Rebuilding and Reorganizing Indexes based on Fragmentation in SQL Server

Why we have to Rebuild and Reorganize the indexes?
As the time passes, data pages are scattered/fragmented and not in order due to insert, update and delete operation. So, its better the rebuild / reorganize the indexes to improve the query performance. Otherwise, query response would be slow because its would take more time to collect the data's which is scattered.

Rebuilding Indexes:
1. Its nothing but dropping and creating the indexes.
2. More CPU resources need.
3. Deadlock would happen while rebuilding the indexes.

Reorganizing Indexes:
1. Rearranging the leaf pages in correct order.
2. Less CPU resources need.
3. Deadlock possibility is very less.

When to use REORGANIZE and REBUILD?
Generally if fragmentation percentage for each index is from 5 - 30 percent, reorganize the index. if fragmentation percentage is more than 30 percent, its better to REBUILD the index.

SQL Script to REORGANIZE and REBUILD each index in each table:

 DECLARE @ReOrgLowPercent INT,  
   @ReOrgHighPercent INT,  
   @TableName VARCHAR(200)  
 SET @ReOrgLowPercent = 5  
 SET @ReOrgHighPercent = 30  
 DECLARE @IndexTable TABLE ( AlterStatement VARCHAR(4000) )  
 DECLARE C CURSOR  
   FOR SELECT name  
     FROM  sys.tables  
 OPEN C  
 FETCH NEXT FROM C INTO @TableName  
 WHILE ( @@FETCH_STATUS = 0 )  
   BEGIN  
     --PRINT @TableName  
     INSERT INTO @IndexTable ( AlterStatement )  
         SELECT CASE when avg_fragmentation_in_percent between @ReOrgLowPercent and @ReOrgHighPercent  
                then 'ALTER INDEX ' + i.name + ' ON ' + t.name  
                  + ' REORGANIZE;'  
                when avg_fragmentation_in_percent > @ReOrgHighPercent  
                then 'ALTER INDEX ' + i.name + ' ON ' + t.name  
                  + ' REBUILD WITH(ONLINE=ON);' --- enterpise version obly  
             END  
         FROM  sys.dm_db_index_physical_stats(DB_ID(),  
                             OBJECT_ID(@TableName),  
                             NULL, NULL, NULL) AS a  
             JOIN sys.indexes AS i ON a.object_id = i.object_id  
                          AND a.index_id = i.index_id  
             JOIN sys.tables AS t ON t.object_id = i.object_id  
     FETCH NEXT FROM C INTO @TableName  
   END  
 CLOSE C  
 DEALLOCATE C  
 SELECT AlterStatement  
 FROM  @IndexTable  
 WHERE  
      AlterStatement IS NOT NULL  

SQL Server 2012 - Improved Version of EXECUTE keyword

In Previous version of SQL Server 2005/2008, if you execute the SELECT Stored Procedure using EXECUTE keyword, it will return all the records as it is. if you want to edit the column name and data type of result set, you have to modify the stored procedure respectively.

But in SQL Server 2012, improved version of EXECUTE keyword allows to modify the result set with changing the column name and data type without modifying the stored procedure.

To explore this features, execute the following script to create a table, insert the record and creating the stored procedure.

 CREATE TABLE Customers  
 (  
      ID INT IDENTITY(1,1),  
      FirstName VARCHAR(50),  
      LastName VARCHAR(50),  
      DOB DATE  
 )  
 GO  
 INSERT INTO Customers (FirstName, LastName, DOB)  
 SELECT 'Ayyanar', 'Jayabalan', '08/12/1982'  
 UNION ALL  
 SELECT 'Greg', 'Damico', '08/12/1970'  
 GO  
 CREATE PROCEDURE Customers_SelectAll  
 AS  
 BEGIN  
      SELECT  
           ID,  
           FirstName,  
           Lastname,  
           DOB  
      FROM dbo.Customers  
 END  
 GO  

In SQL Server 2005/2008:

 -- SQL 2005/2008  
 EXECUTE Customers_SelectAll  

RESULT:


In SQL Server 2012:
Here i am changing the column "id" into "CustomerId" and also changing the data type of firstname from VARCHAR(50) to VARCHAR(4). Since i reduced the length of FirstName to 4, it shows truncated firstname value.

 -- SQL 2012  
 EXEC Customers_SelectAll  
 WITH RESULT SETS  
 (  
 (  
      CustomerID SMALLINT,  
      FirstName VARCHAR(4),  
      Lastname VARCHAR(20),  
      DateOfBirth DATE  
 )  
 );  

RESULT:

SQL Server 2012 - Throwing Exception using THROW in TRY CATCH BLOCK instead of using RAISERROR

In SQL Server 2005/2008, if you want to throw an error again in catch block of TRY CATCH statement, you have to use RAISERROR with ERROR_MESSAGE(), ERROR_SEVERITY().

In SQL Server 2005/2008:

 -- EXCEPTION HANDLING in SQL Server 2005/2008  
 BEGIN TRY  
   DECLARE @VALUE INT  
   SET @VALUE = 1 / 0  
 END TRY   
 BEGIN CATCH  
   DECLARE @Message nvarchar(4000), @Severity int  
   SELECT @Message = ERROR_MESSAGE(), @Severity = ERROR_SEVERITY()  
   RAISERROR ( @Message, @Severity, 1 )  
 END CATCH  

SQL Server 2012:
Just, you can use THROW keyword to throw the complete error information.

 -- EXCEPTION HANDLING using THROW in SQL Server 2012  
 BEGIN TRY  
      DECLARE @VALUE INT  
      SET @VALUE = 1/0  
 END TRY   
 BEGIN CATCH  
      THROW  
 END CATCH  

if you execute the above statement, you will receive an error like below.

 Msg 8134, Level 16, State 1, Line 5  
 Divide by zero error encountered.  

Custom Paging in SQL Server 2012 - Using OFFSET

In SQL Server 2005/2008, if you have want to implement the custom paging for ASP.Net GridView or any other controls, you have use either COMMON TABLE EXPRESSIONS with ROW_NUMBER() or just ROW_NUMBER() keyword.

CUSTOM PAGING IN SQL SERVER 2005/2008:

 -- CUSTOM PAGING IN SQL SERVER 2005   
 USE AdventureWorks  
 DECLARE @PageIndex SMALLINT, @PageSize SMALLINT, @StartRowId INT, @EndRowId INT  
 SET @PageIndex = 1  
 SET @PageSize = 10  
 SET @StartRowId = ( (@PageIndex - 1) * @PageSize) + 1  
 SET @EndRowId = (@StartRowId + @PageSize) - 1  
 SELECT      *  
 FROM (  
 SELECT ROW_NUMBER() OVER(ORDER BY EmployeeId) AS RowId, *  
 FROM HumanResources.Employee) AS TT  
 WHERE RowId >= @StartRowId and RowId <= @EndRowId  

CUSTOM PAGING IN SQL SERVER 2012 using OFFSET:
In SQL Server 2012, by using the OFFSET feature, we can implement the custom paging easily without having too many SQL statements.
Just you have to specify starting row id and page size like below.

 OFFSET <Offset> ROWS  
 FETCH NEXT <PageSize> ROWS ONLY;  

Here is the complete T-SQL statement for custom paging.

 -- CUSTOM PAGING IN SQL SERVER 2012 RC0  
 DECLARE @PageIndex SMALLINT, @PageSize SMALLINT, @Offset INT  
 SET @PageIndex = 1  
 SET @PageSize = 10  
 SET @Offset = ( (@PageIndex - 1) * @PageSize)  
 SELECT      *  
 FROM HumanResources.Employee  
 ORDER BY EmployeeId  
 OFFSET @Offset ROWS  
 FETCH NEXT @PageSize ROWS ONLY;  

SQL Server 2012 Features - Sequence Number

Today,i have installed to explore the features of SQL Server 2012. First feature which i am going to discuss is "Sequence Number". You can generate sequence number like identity column. In Previous release, if you want to generate a sequence id before inserting into the table, you have to create a seperate physical table with identity column. Before inserting record into table, we have to insert a row in custom sequence table and get the latest value.

But in SQL Server 2012, you no need to insert record in any physical custom table to retrieve the number in sequence. Just you have to create sequence object which binds to schema.

Here are the steps to create Sequence object using SQL Server Management Studio.

1. Create a New Sequence:



2. Specify the NAME, MIN, MAX, CYCLE and OTHER PROPERTIES:



3. Once you entered the values for sequence object. You have to use following query to get NEXT sequence number using NEXT VALUE FOR keyword.

 select NEXT VALUE FOR dbo.GlobalSequence;  

RESULT:

 1  

If you execute above query, you will get the value as 1. Next time, if you executing same query, you will get the value as 2. Same way, you would get the number in sequential for every execution.

4. With CYCLE OPTION:

 select NEXT VALUE FOR dbo.GlobalSequence;  
 select NEXT VALUE FOR dbo.GlobalSequence;  
 select NEXT VALUE FOR dbo.GlobalSequence;  
 select NEXT VALUE FOR dbo.GlobalSequence;  
 select NEXT VALUE FOR dbo.GlobalSequence;  
 select NEXT VALUE FOR dbo.GlobalSequence;  

if you execute the above query, you would get values like below.

 1  
 2  
 3  
 4  
 5  
 1  

if sequence value reaches maximum value, it will start again with start value if you enabled the CYCLE option. Otherwise, it will throw an error like below;

 The sequence object 'GlobalSequence' has reached its minimum or maximum value. Restart the sequence object to allow new values to be generated.  

SQL Server 2012 - Testimonials

http://www.microsoft.com/sqlserver/en/us/future-editions.aspx

SQL SERVER – SHRINK Log File using SHRINKFILE

if you want to shrink the log file, you would use SHRINKFILE command like.

DBCC SHRINKFILE(TestDBLog, 1)
GO

But in some cases (like production database), if you shrink log file by using above command, you won't get expected result. So, if you want to shrink log file to minimum size, take backup (*.bak) of entire database with SIMPLE recovery mode and then run the shrink file command.

Here is the code to backup the database and shrink the log file

BACKUP DATABASE [TestDb] TO DISK = N'C:\TestDb.bak'
GO
DBCC SHRINKFILE(TestDBLog, 1)
GO