Monday 23 December 2013

Error: Index was outside the bounds of the array

It is said that we may work on SQL Server connecting to Object Explorer by using the SQL Management Studio tool (SSMS) of an older version. It is true up to a point, but there is a risk of getting some unexpected errors, for instance, we might get this error while working connected to SQL Server 2012 Object Explorer by using SQL Server 2005 SSMS.



More specifically, the error is raised because superior versions support more features, which means that each feature is likely to have one folder on the Object Explorer. Put differently, there is a bigger number of folders to display in SQL Server 2012 than SQL Server 2005 whose Object Explorer cannot display more because the max array index is outside the bounds that this supports. Ideally, it is highly advisable to always use the latest SSMS to connect to older versions (but not vice versa). As simple as water. That is all for now, let me know any remarks you may have. Thanks for reading.

Sunday 12 December 2010

SQL11 (Denali): New object SEQUENCE

SQL Server 2012 CTP1 is coming up with a new object named SEQUENCE which allows to manage sequence numbers between 2^31 – 1 and 2^31 –1. SEQUENCE solves many problems with respect to using the IDENTITY property because it is not tied to the column and can be used for many columns in different tables. In this capacity, this novelty is useful as alternative solution and in replacement of ROW_NUMBER for some cases. SEQUENCE can be used with UNION ALL but not with Functions, DISTINCT, UNION, EXCEPT and INTERSECT. The way of using is very easy, practical, flexible and  offers better performance than IDENTITY. Creating the right indexes on columns which takes values from SEQUENCE object will help tremendously, indeed.

Creating a SEQUENCE object is simple.

CREATE SEQUENCE dbo.MySeq
    AS INT 
    START WITH 1
    INCREMENT BY 1
    MINVALUE 1
    MAXVALUE 100
    CYCLE 
    CACHE 20
;
Now using it:
SELECT NEXT VALUE FOR dbo.MySeq
GO 4
The result would be something like this:

(no column name)
1
2
3
4
If we execute again the previous query, the result will be numbers from 5 to 8, and so on. However, if we want to reinitialise the sequence, it can be done as follows:

ALTER SEQUENCE dbo.MySeq RESTART 
Now using SEQUENCE as an alternative of ROW_NUMBER:
select next value for dbo.MySeq as [nro],  Employees.FirstName   from Employees 
image
And also in this way:

select next value for dbo.MySeq  over (order by customers.CustomerID ) as [nro], 
       customers.CustomerID,  Customers.ContactName 
 from Customers 
Here are some resources to check out more information about SEQUENCE:
That is all for now, let me know any remark you may have. Thanks for reading.

Friday 12 November 2010

SQL11 (Denali): OFFSET and FETCH (New options for ORDER BY)

It is known that ORDER BY clause has been improved in 'Denali' (code name of the upcoming version SQL Server 2012) adding more options as a result from looking for new solutions to work on specific business cases such as fetching top N rows per window of results from the result set which we used to do by combining TOP, OVER, and ORDER BY. Nevertheless, the performance of this approach was poor and not even close to being as good as OFFSET and FETCH in 'Denali'.

OFFSET keyword is used for excluding first N rows (default value is zero) whereas FETCH is used for reading/fetching M rows per each window of results starting from position (N+1). Therefore, OFFSET and FETCH allow to get only rows from position (N+1) to (N+M). I do recommend creating one Index (preferably Index Clustered) on ORDER BY columns and using these columns on WHERE clause as much as possible in order to avoid costly operations such as Index Scans or Table Scans. I have performed many tests on 10 millions of rows, and to be to honest, it worked very well. To illustrate, the following query will start fetching from position 4+1=5 (OFFSET position=4 for SQL Server) to the end of the table.

SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address   
FROM dbo.Customers    
ORDER BY CustomerID 
   OFFSET (4) ROWS

Now with DESC the results will be different (excluding last 4 rows)

SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address   
FROM dbo.Customers    
ORDER BY CustomerID  DESC
   OFFSET (4) ROWS

Basically, each execution of the query is done independently, which means that not only are the results sent back to the client side, but also immediately the resources are released. Here is another example how to get only the top 3 rows excluding first 4 rows (OFFSET=4), which means that it only reads the fifth, sixth, and seventh rows.

SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
FROM dbo.Customers    
ORDER BY CustomerID  
   OFFSET (4) ROWS 
   FETCH NEXT 3 ROWS ONLY --  FETCH FIRST 3 ROWS ONLY
GO

Having a need to read the next three rows (eighth, ninth, and tenth rows), we have to increase OFFSET value in three.
 
SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
FROM dbo.Customers    
ORDER BY CustomerID  
   OFFSET (7) ROWS ---ó OFFSET (7) ROW 
   FETCH NEXT 3 ROWS ONLY -- ó FETCH FIRST 3 ROWS ONLY
GO


Here the script to read dynamically each result by using variables (and optionally OPTIMIZE FOR hint to optimise this Ad-Hoc query). 
DECLARE @Start INT, @Next INT
SET @Start=0 

SET @Next=3
SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
FROM dbo.Customers    
ORDER BY CustomerID 
     OFFSET (@Start) ROWS    
     FETCH NEXT @Next ROWS ONLY 

 OPTION(OPTIMIZE FOR (@Start=4,@Next=5))
GO
Now the same code inside an stored procedure.

CREATE PROCEDURE dbo.PageFetch(@PageNumber INT, @PageSize INT)
AS
BEGIN
SELECT ROW_NUMBER() OVER (ORDER BY CustomerID) AS 'Nro', 
CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
FROM dbo.Customers    
ORDER BY CustomerID 
     OFFSET (@PageSize * (@PageNumber -1) ) ROWS
     FETCH NEXT @PageSize ROWS ONLY 
END

Querying the results for second, third and fifth windows of results with OFFSET.
EXEC dbo.PageFetch 2,3;
GO
EXEC dbo.PageFetch 3,3;
GO
EXEC dbo.PageFetch 5,3;


Luckily, OFFSET and FETCH are supported by subqueries, functions, derived tables, but not by indexed views and using directly with TOP, OVER, INSERT, UPDATE, MERGE, o DELETE (except inside of independent queries at lower level such as subqueries). Here an illustration.

SELECT  TOP(2) ROW_NUMBER() OVER(ORDER BY CustomerID) AS 'NRO', CompanyName, ContactName, ContactTitle  FROM (
        SELECT  CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
        FROM dbo.Customers    
        ORDER BY CustomerID  
           OFFSET (4) ROWS ---ó OFFSET (4) ROW 
           FETCH NEXT 3 ROWS ONLY -- ó FETCH FIRST 3 ROWS ONLY
) AS T1

Finally, using it with Views.

CREATE VIEW dbo.v1
as
SELECT CustomerID, CompanyName, ContactName, ContactTitle, Address --/..*.../  
FROM dbo.Customers  
ORDER BY CustomerID 
   OFFSET (0) ROWS ---ó OFFSET (4) ROW 
GO
SELECT CustomerID, CompanyName, ContactName FROM dbo.v1
WHERE  CustomerID LIKE 'AN%' 

To sum up, OFFSET and FETCH are excellent improvements which allow to implement solutions to fetch only specific pages/windows of results from the result set with ease. Let me know any remarks you may have. Thanks for reading!

Monday 18 October 2010

SQL11 (Denali): How to migrate SQL Server 2000 databases to SQL Server 2012?

Having found 'SQL Server 2000 (80)' inside the compatibility levels supported by SQL Server 2012, I need to say that the upcoming version of SQL Server is not going to support it. we do know that each new version only supports compatibility levels of two below versions, which means that it is an error as this is a CTP build.



What is going on if we try to restore a SQL Server 2000 database on SQL Server 2012? well, we will get the following error:

Msg 3169, Level 16, State 1, Line 1
The database was backed up on a server running version 8.00.0760. That version is incompatible with this server, which is running version 11.00.1103. Either restore the database on a server that supports the backup, or use a backup that is compatible with this server.
Msg 3013, Level 16, State 1, Line 1
RESTORE DATABASE is terminating abnormally.


Attaching the database neither works.

Msg 1813, Level 16, State 2, Line 1
Could not open new database ‘Northwind’. CREATE DATABASE is aborted.
Msg 1813, Level 16, State 2, Line 1
Could not open new database ‘Northwind’. CREATE DATABASE is aborted.
Msg 950, Level 20, State 1, Line 1
Database ‘Northwind’ cannot be upgraded because its non-release version (0) is not supported by this version of SQL Server. You cannot open a database that is incompatible with this version of sqlservr.exe. You must re-create the database.


Therefore, SQL Server 2000 databases have to be migrated to SQL Server 2012 following the next steps: (1) restore SQL Server 2000 databases on SQL Server 2005 or SQL Server 2008; (2) take Full Backups of the new SQL Server 2005/2008 databases, and finally; (3) restore them on SQL Server 2012. In other words, there will not be direct migration from SQL Server 2000 to SQL Server 2012. That is all for now, let me know any remarks you may have. Thanks for reading again!
HELLO, I'M PERCY REYES! — a book lover, healthy lifestyle lover... I've been working as a senior SQL Server Database Administrator (DBA) for over 20 years; I'm a three-time awarded Microsoft Data Platform MVP. I'm currently doing a PhD in Computer Science (cryptography) at Loughborough University, England — working on cryptographic Boolean functions, algorithmic cryptanalysis, number theory, and other algebraic aspects of cryptography. READ MORE