Dataset

Enhanced Definition

A collection of related data records stored as a unit on mainframe storage, equivalent to a file in other operating systems. Datasets are the fundamental unit of data organization in z/OS.

Key Characteristics

- Named collection of records
- Defined organization and structure
- Stored on DASD or tape
- Managed through catalog system
- Protected by security software
Dataset Organizations:
- PS (Physical Sequential): Sequential records, simple structure
- PO (Partitioned): Library containing multiple members
- VSAM (Virtual Storage Access Method): Advanced indexed/sequential
- KSDS (Key-Sequenced Dataset)
- ESDS (Entry-Sequenced Dataset)
- RRDS (Relative Record Dataset)
- Linear (byte-stream for specialty uses)
- DA (Direct Access): Direct organization datasets
- IS (Indexed Sequential): Legacy indexed organization
Dataset Naming:
- Up to 44 characters total length
- Qualifiers separated by periods
- Each qualifier 1-8 alphanumeric characters
- First character must be alphabetic or national
- Example: USER.PROJECT.SOURCE.COBOL
Dataset Attributes:
- RECFM (Record Format): F, FB, V, VB, U
- LRECL (Logical Record Length): Record size in bytes
- BLKSIZE (Block Size): Physical block size
- DSORG (Dataset Organization): PS, PO, VS
- Space Allocation: Primary and secondary extents
Dataset Types by Use:
- Source Code: Program source files
- Object Code: Compiled programs
- Load Libraries: Executable programs
- JCL Libraries: Job control language
- Data Files: Application data
- System Libraries: OS components

Use Cases

- Storing application programs and data
- Organizing system software components
- Managing source code repositories
- Maintaining production data files
- Archiving historical information