Indexes include a subset of the columns of one or more tables. Typically, indexes include the column values from all rows in the table. Sparse join indexes further limit the index to include only a subset of the table rows, in addition to a subset of table columns. Sparse JIs can include one or more tables, and can also be aggregate JIs.
If queries frequently are limited to a subset of the rows of a table, for example, rows with a specific value in one column, a sparse JI that includes only those rows can improve performance by providing a more limited data set to be queried.