Numeric data types

For storing numerical data.

The NUMERIC and DECIMAL data types, and the various kinds of INTEGER data types, are sometimes called exact numeric data types, in contrast to the approximate numeric data types FLOAT, DOUBLE, and REAL.

The exact numeric data types are those for which precision and scale values can be specified, while approximate numeric data types are stored in a predefined manner. Only exact numeric data is guaranteed accurate to the least significant digit specified after an arithmetic operation .
Before release 5.5, hexadecimal constants longer than four bytes were treated as string constants, and others were treated as integers. The new default behavior is to treat them as binary type constants. To use the historical behavior, set the TSQL_HEX_CONSTANTS database option to OFF.
Data type lengths and precision of less than one are not allowed.

Only the NUMERIC data type with scale = 0 can be used for the Transact-SQL identity column.
You should avoid default precision and scale settings for NUMERIC and DECIMAL data types, because these are different between Adaptive Server Anywhere and Adaptive Server Enterprise. In Adaptive Server Anywhere, the default precision is 30 and the default scale is 6. In Adaptive Server Enterprise, the default precision is 18 and the default scale is 0.
The FLOAT ( p ) data type is a synonym for REAL or DOUBLE, depending on the value of p. For Adaptive Server Enterprise, REAL is used for p less than or equal to 15, and DOUBLE for p greater than 15. For Adaptive Server Anywhere, the cutoff is platform-dependent, but on all platforms the cutoff value is greater than 15.

For information about changing the defaults by setting database options, see PRECISION option [database] and SCALE option [database].