Beginning Microsoft SQL Server 2008 ... - S3 Tech Training

Chapter 6: Constraints
At a more specific level, we’ll be looking at the methods of implementing each of these types of constraints,
including:
❑ Primary key constraints
❑ Foreign key constraints
❑ Unique constraints (also known as alternate keys)
❑ Check constraints
❑ Default constraints
❑ Rules
❑ Defaults (similar to default constraints)
We’ll also take a very cursory look at triggers and stored procedures (there will be much more on these
later) as methods of implementing data integrity rules.
Types of Constraints
There are a number of different ways to implement constraints, but each of them falls into one of three
categories — domain, entity, or referential integrity constraints, as illustrated in Figure 6-1.
Domain Constraints
Domain constraints deal with one or more columns. What we’re talking about here is ensuring that a particular
column or set of columns meets particular criteria. When you insert or update a row, the constraint
is applied without respect to any other row in the table — it’s the column’s data you’re interested in.
For example, if we want to confine the UnitPrice column to include only values that are greater than or
equal to zero, that would be a domain constraint. Although any row that had a UnitPrice that didn’t
meet the constraint would be rejected, we’re actually enforcing integrity to make sure that the entire column
(no matter how many rows) meets the constraint. The domain is the column, and our constraint is a
domain constraint.
We’ll see this kind of constraint when dealing with check constraints, rules, defaults, and default constraints.
Entity Constraints
154
Entity constraints are all about individual rows. This form of constraint doesn’t really care about a column
as a whole; it’s interested in a particular row, and would best be exemplified by a constraint that
requires every row to have a unique value for a column or combination of columns.
“What,” you say, “a unique column? Doesn’t that mean it’s a domain constraint?” No, it doesn’t. We’re
not saying that a column has to meet any particular format, or that the value has to be greater or less than
anything. What we’re saying is that for this row, the same value can’t already exist in some other row.
We’ll see this kind of constraint when dealing with primary key and unique constraints.