The Advanced JDBC Column handler optionally writes character data (CLOB, LONGVARCHAR types) to the file system, but by default these character values are returned inline as the Guided Search property’s value. If configured for output to the file system using the charDataToDisk option mentioned above, the files will be created in the outputDataDir directory (also configurable) and would have filenames of the form clob_dataN.tmp. N, in this case, is a random number suffix to keep these temporary files distinct.

The binary column type handlers always write their data to the file system, in the outputDataDir directory. These file names are of the form blob_dataN.tmp. The relative path to each file is returned as the Guided Search property’s value. For example ../incoming/blob_data22607.tmp. The pipeline must then read this file in a subsequent record manipulator.