Comparison of the VARCHAR(max) and VARCHAR(n) SQL Server Data Types
By: Daniel Farina | Updated: 2016-10-06 | Comments (5) | Related: 1 | 2 | 3 | More > Data Types
You are developing a 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 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 the this data type. Read this tip to learn what you should know about the SQL Server VARCHAR(Max) vs. VARCHAR(n) data types.
In order to decide between using VARCHAR(Max) or VARCHAR(n) 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) 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 text. 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 tip in case you need a further explanation on the differences between VARCHAR 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).
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:
- LTRIM and RTRIM
- UPPER and LOWER
- LEFT and RIGHT
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)
The value that is copied to Lob_column_name. If expression is set to NULL, Length is ignored, and the value in Lob_column_name is truncated at the specified Offset.
The starting position in the value of Lob_column_name at which expression is written. If Offset is NULL, the update operation appends expression at the end of Lob_column_name ignoring the value of Length parameter.
The length of the section in the column, starting from Offset, which is replaced by expression. If Length is NULL, the update operation removes all data from Offset to the end of the Lob_column_name value.
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.
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.
- 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.
Last Updated: 2016-10-06
About the author
View all my tips