Types of data sets

z/OS® has many different types of data sets.

Data sets can be sequential or partitioned:

In a sequential data set, records are data items that are stored consecutively.
A partitioned data set consists of a directory and members. The directory holds the address of each member and thus makes it possible to access each member directly. Each member consists of sequentially stored records.

Partitioned data sets are often called libraries. By convention, libraries often have the letters "LIB" in the data set name. Also by convention, programs and procedures are stored in separate libraries; within a library, each program or procedure is stored as a separate member of the partitioned data set.

Data sets can be permanent or temporary:

Most permanent data sets exist before a job starts and persist after a job step completes. Some permanent data sets are created during a job step and persist after the job completes.
Temporary data sets generally are used to pass data from one job step to another, and exist only during the life cycle of the job.

Data sets can be cataloged, which permits the data set to be referred to by name without specifying where the data set is stored. A catalog describes data set attributes and indicates the devices on which a data set is located. In z/OS, the master catalog and user catalogs store the locations of data sets.