When you create an index, a histogram is created on the first column of the index. The histogram stores information about the distribution of values in the column. Then you can use update statistics to generate statistics for the minor keys of a compound index and columns used in unindexed search clauses.
The histogram for a column contains data in a set of steps or cells. You can specify the number of cells can when the index is created or when the update statistics command is run. For each cell, the histogram stores a column value and a weight for the cell.
There are two types of cells in histograms:
A frequency cell represents a value that has a high proportion of duplicates in the column. The weight of a frequency cell times the number of rows in the table equals the number of rows in the table that match the value for the cell. If a column does not have highly duplicated values, there are only range cells in the histogram.
Range cells represent a range of values. Range cell weights and the range cell density are used for estimating the number of rows to be returned when search argument values falls within a range cell.
For more information on histograms, see “Histogram displays”.