Summary

Tables are the heart of relational databases in general and SQL Server in particular. In this chapter, we've looked at datatype issues, including size and performance tradeoffs. SQL Server provides a datatype for almost every use. We also looked at variable-length datatype issues and saw that it's simplistic to think that variable-length datatypes are either always good or always bad to use. You've also seen how data is physically stored in data pages, and we discussed some of the system table entries that are made to support tables.

SQL Server provides user-defined datatypes for support of domains, and it provides the IDENTITY property to make a column produce autosequenced numeric values. It also provides constraints, which offer a powerful way to ensure your data's logical integrity. In addition to the NULL/NOT NULL designation, SQL Server provides PRIMARY KEY, UNIQUE, FOREIGN KEY, CHECK, and DEFAULT constraints. Our discussion of these features also touched on a number of pragmatic issues, such as performance implications, conflicts between different constraints and triggers, and transaction semantics when a constraint fails.