Available as PDF
A Partitioned Data Set (PDS) is an IBM OS/390 (MVS) file that contains one or more sequential files called members. Members can be used to store various types of information:
General alphanumeric data, such as the contents of database records and fields.
Text files, such as paragraphs or chapters in a novel.
Source files, which are uncompiled job-specific instructions written in a programming language like Fortran, SAS, C, or COBOL.
Object decks, which are compiled programs, not yet link-edited.
Macros, a series of computer instructions invoked during the translation of a single coded word.
Executable programs, stored as machine language, produced by compiling and link-editing a source file. An example would be a subroutine program. A PDS whose members are independently executable is identified in this document as a load module library. The synonyms program library and load library are also used here.
Batch job instructions, using Job Control Language (JCL).
TSO CLISTs and REXX execs.
Partitioned data sets are versatile. Information in one PDS member can be accessed quickly, directly, without disturbing adjacent members. Members are located via a directory, where each has a unique entry. At the same time, PDS format allows processing of sequential members in batch operations.
Created PDSs should be compact to save on storage costs and conserve disk space, yet flexible enough to permit growth in the data set. For example, if you don't allocate enough directory blocks when you create a PDS you may have to subsequently create another, larger PDS and transfer all your data to it. Conversely, allocating more directory blocks than will ever be used is wasteful and can result in unnecessary disk storage charges. For more information on directories and other PDS parameters, see Chapter 2 of this document, How to Create a Partitioned Data Set.
A PDS may be stored on tape, in an unloaded format, or on a direct access (disk drive) volume. However, data within a PDS stored on tape can only be accessed or processed after the PDS has been reloaded onto a disk drive volume.
A PDS should be created with a data set name conforming to EI&O naming conventions, under which it will be cataloged. The names of its individual members must be one to eight characters in length, and unique within that data set. However, after a PDS is created, its members can be assigned one or more aliases, which are alternate names for the same member. Aliases permit access to information under familiar names. For further details on EI&O naming conventions, see EI&O Document D0045.