Guru's Guide to SQL Server Architecture and Internals, The

Overview

A cursor is a mechanism for accessing the rows in a table or result set on a piecemeal basis—one at a time. They run counter to SQL Server's normal way of doing things by parceling result sets into individual rows; fetching a row from a cursor is analogous to returning a single row via a SELECT statement. Unlike a traditional result set, a cursor keeps track of its position automatically and provides a wealth of facilities for scrolling around in the underlying result set. Cursors also provide a handy means of updating the underlying result set in a positional fashion and of returning result set pointers via variables.

The advice I usually give people who are thinking about using cursors is not to. If you can solve a problem using Transact-SQL's ...