Comparison of the VARCHAR(max) and VARCHAR(n) SQL Server Data Types

Problem

You are developing a Microsoft SQL Server application and at some point in your code you receive the error “String or binary data would be truncated”. After digging into the T-SQL code you discovered that this error message is caused because a table has a VARCHAR(1000) column and you need to insert more than 1000 characters. Someone on your team suggest using the VARCHAR(max) data type, but you are unsure about the implications of this character data type.  Read this tip to learn what you should know about the SQL Server VARCHAR(Max) vs. VARCHAR(n) data types.

Solution

In order to decide between using VARCHAR(Max) or VARCHAR(n) variable length data types we have to understand that even when both data types look the same, they have some major differences pertaining to storage, behavior and of course their intended usage.

The VARCHAR(Max) SQL Server Data Type

The VARCHAR(Max) as well as NVARCHAR(max) and VARBINARY(max) string data types were first introduced in SQL Server 2005 to replace the large object (LOB) data types TEXT, NTEXT and IMAGE respectively. All of these data types can store up to 2GB of data except NVARCHAR(max) that can store 1GB of Unicode characters. As you may guess, the maximum storage for these data types is the same as the ones being replaced. I suggest reading the SQL Server differences of char, nchar, varchar and nvarchar data types tutorial in case you need a further explanation on the differences between VARCHAR data types and NVARCHAR data types.

A common question amongst beginners is: if the VARCHAR(max) data type can store up to 2GB why can’t we declare a column or variable of type VARCHAR(10000)? Why are we limited to either declare a VARCHAR(8000) or VARCHAR(max)? The reason behind this is that VARCHAR(n) is intended to store the data in the row page. SQL Server stores data in 8KB pages (8,192 bytes), but the page header occupies the first 96 bytes of each data page leaving 8,096 bytes for data, row overhead, and row offsets, that’s why you cannot declare a value greater than VARCHAR(8000).

Advantages

One of the advantages of the VARCHAR(max) data type as a replacement of TEXT data type is that we can declare local variables to manipulate LOBs and even declare VARCHAR(max) parameters on functions and stored procedures. This is something that cannot be done with the TEXT data type. Furthermore, the VARCHAR(max) data type can be used inside string functions such as REPLACE, CHARINDEX or LEFT instead of using READTEXT, WRITETEXT, and UPDATETEXT commands to manipulate LOBs.

The following list includes the functions that support the VARCHAR(max) data type:

• STRING_SPLIT
• STRING_ESCAPE
• CONCAT
• PATINDEX
• CHARINDEX
• LTRIM and RTRIM
• UPPER and LOWER
• LEN
• LEFT and RIGHT
• REPLACE
• REPLICATE
• REVERSE
• SUBSTRING
• STUFF

Updating Data on VARCHAR(max) Columns

Something that not many people know is that the UPDATE statement allows us to append and replace data in VARCHAR(max), NVARCHAR(max) and VARBINARY(max) columns without using string functions. There is an implicit WRITE function in the UPDATE statement for columns of VARCHAR(max), NVARCHAR(max) and VARBINARY(max) data type with the following syntax.

UPDATE MyTable SET
Lob_column_name.WRITE (expression,Offset,Length)

Limitations

There is an obvious limitation on VARCHAR(max) columns which is that these columns cannot be indexed.  If you think you need to place an index on a VARCHAR(max) column I suggest that you may need to review the table design.

Sample Usage

Let’s create a new database to do some testing.

USE [Employees];
GO
CREATE TABLE Sellers
    (
      SellerID INT IDENTITY(1, 1)
                   PRIMARY KEY ,
      Name VARCHAR(255) ,
      Percentage NUMERIC(3, 2) ,
      Commissions  MONEY
    )
GO
INSERT INTO dbo.Sellers
SELECT 'John Doe', 0.25, 0
UNION ALL
SELECT 'Linda Smith', 0.1, 0
GO

On the Sales database create a table named Transactions to store sales information.

USE [master]
GO
CREATE DATABASE [TestDB]
 CONTAINMENT = NONE
 ON  PRIMARY 
( NAME = N'TestDB', FILENAME = N'E:\MSSQL\TestDB.mdf' ,
 SIZE = 10MB , MAXSIZE = UNLIMITED, FILEGROWTH = 1024KB )
 LOG ON 
( NAME = N'TestDB_log', FILENAME = N'E:\MSSQL\TestDB_log.ldf',
 SIZE = 10240KB , MAXSIZE = 2048GB , FILEGROWTH = 10%)
GO

The following script will create a test table with one VARCHAR(Max) column and an identity column, and then insert one row with 1,000,000 characters.

USE [TestDB]
GO
CREATE TABLE TestTable
    (
      ID INT IDENTITY(1, 1) ,
      TestData VARCHAR(MAX) ,
      PRIMARY KEY CLUSTERED ( ID )
    )
GO
DECLARE @str VARCHAR(MAX) = 'a'
INSERT  INTO dbo.TestTable
        ( TestData )
        SELECT  REPLICATE(@str, 1000000)
GO

Now we will append the text “0123” at the end of the string.

USE [TestDB]
GO
UPDATE  dbo.TestTable
SET     TestData.WRITE('0123', NULL, NULL)
WHERE   ID = 1;
SELECT  RIGHT(testdata, 10) ,
        LEN(TestData)
FROM    dbo.TestTable;
GO

As you can see on the next image the data was appended and the string length was increased.

Next Steps

• Since TEXT, NTEXT and IMAGE data types will be no longer supported; you can use this tip to Identify All SQL Server Tables with Columns of a BLOB Data Type to upgrade to the supported data types.
• In this tip you will find a more detailed explanation about the maximum row size in a SQL Server table: Row sizes exceeding 8060 bytes in Sql 2005.
• Check out the SQL Server Data Types Tips Category for more tips.