Difference between different string types in SQL Server?

Question

What is the difference between char, nchar, ntext, nvarchar, text and varchar in SQL?

I

Answer 1

The n prefix simply means Unicode. They "n" types work similarly to the plain versions except they work with Unicode text.

char is a fixed length field. Thus char(10) filled with "Yes" will still take 10 bytes of storage.

varchar is a variable length field. char(10) filled with "Yes" will take 5 bytes of storage (there is a 2 byte overhead for using var data types).

char(n) holding string of length x. Storage = n bytes. varchar(n) holding string of length x. Storage = x+2 bytes.

vchar and nvarchar are similar except it is 2 bytes per character.

Generally speaking you should only use char & char (over varchar & nvarchar) when working with fixed or semi-fixed strings. A good example would be a product_code or user_type which is always n characters long.

You shouldn't use text (or ntext) as it has been deprecated. varchar(max) & nvarchar(max) provides the same functionality.

Answer 2

text and ntext are deprecated, so lets omit them for a moment. For what is left, there are 3 dimensions:

• Unicode (UCS-2) vs. non-unicode: N in front of the name denotes Unicode
• Fixed length vs. variable length: var denotes variable, otherwise fixed
• In-row vs. BLOB: (max) as length denotes a BLOB, otherwise is an in-row value

So with this, you can read any type's meaning:

• CHAR(10): is an in-row fixed length non-Unicode of size 10
• NVARCHAR(256): is an in-row variable length Unicode of size up-to 256
• VARCHAR(MAX): is a BLOB variable length non-Unicode

The deprecated types text and ntext correspond to the new types varchar(max) and nvarchar(max) respectively.

When you go to details, the meaning of in-row vs. BLOB blurs for small lengths as the engine may optimize the storage and pull a BLOB in-row or push an in-row value into the 'small BLOB' allocation unit, but this is just an implementation detail. See Table and Index Organization.

From a programming point of view, all types: CHAR, VARCHAR, NCHAR, NVARCHAR, VARCHAR(MAX) and NVARCHAR(MAX), support an uniform string API: String Functions. The old, deprecated, types TEXT and NTEXT do not support this API, they have a separate, deperated, TEXT API to manipulate. You should not use the deprecated types.

BLOB types support efficient in-place updates by using the UPDATE table SET column.WRITE(@value, @offset) syntax.

The difference between fixed-length and variable length types vanishes when row-compression on a table. With row-compression enabled, fixed lenght types and variable length are stored in the same format and trailing spaces are not stored on disk, see Row Compression Implementation. Note that page-compression implies row-compression.

Answer 3

n-prefix: unicode. var*: variable length, the rest is fixed length.

All data types are properly and nicely... documented.

Like here:

http://msdn.microsoft.com/en-us/library/ms187752.aspx

Is there really an application case for each of these types, or are some of them just deprecated?

No, there is a good case for ANY of them.