In IBM JCL, a concatenation of datasets can contain sequential datasets and DD entries for input, and a dummy dataset can be placed in the concatenation to signify the end of input or empty input. Concatenated datasets cannot be used for writing multiple outputs.