T-SQL Tip 12: Default Clustered Index

We talked about why you should use a clustered index in T-SQL Tip 4. Now let’s talk about what to make the clustered index. Here’s an acronym for things that make a good clustered index:

• One Column: Small data footprint, generally not strings, one column
• Unique: Something that by design won’t repeat
• Increasing. You won’t put a row in the middle of what already exists
• Immutable: Unchanging, and not going to be updated

It’s for those characteristics that I generally recommend an int or bigint IDENTITY column.