What is Data Compression?

In SQL SERVE 2008 Data Compression comes in two flavors:

Row Compression
Page Compression
Row Compression

Row compression changes the format of physical storage of data. It minimize the metadata (column information, length, offsets etc) associated with each record. Numeric data types and fixed length strings are stored in variable-length storage format, just like Varchar.

Page Compression

Page compression allows common data to be shared between rows for a given page. Its uses the following techniques to compress data:

Row compression.
Prefix Compression. For every column in a page duplicate prefixes are identified. These prefixes are saved in compression information headers (CI) which resides after page header. A reference number is assigned to these prefixes and that reference number is replaced where ever those prefixes are being used.
Dictionary Compression.

Dictionary compression searches for duplicate values throughout the page and stores them in CI. The main difference between prefix and dictionary compression is that prefix is only restricted to one column while dictionary is applicable to the complete page.