In the past, database administrators managing tables containing millions of rows of data were forced to create multiple tables. Once those tables were partitioned, the administrator had to tie the tables back together during the execution of many of their queries. Tying partitions together involved creating a partitioned view or a wrapper stored procedure that figured out where the data lived and executed another stored procedure to hit only the partitions needed to return the dataset. The Enterprise and Developer editions of SQL Server™ 2005 let you partition large amounts of data contained in a single table into multiple smaller partitions that can be managed and maintained more effectively. The ability to create data segments that are accessed through a single point-of-entry reduces many of the administration issues that came with the old way of doing things. Using a single point of entry (table name or index name) hides the multiple data segments from the application code and allows the administrator or developer to change the partitions as necessary without having to adjust the code base.
Continue reading here