MVS- Datasets

A data set is the mainframe word for a file. Almost everything on MVS - programs, JCL, reports, databases - is stored in data sets.

This page explains how data sets are named, how they are organized, what RECFM / LRECL / BLKSIZE mean, and how space is allocated on disk.

What is a data set

• A data set is a collection of logically related data records stored on one disk volume or a set of volumes.
• It can be a source program, a library of macros, or a file of records used by a processing program.
• The basic unit of information is the logical record.
• Most MVS data processing is record-oriented. Byte-stream files, as on PCs, are not the traditional style.
• Data is stored on DASD (disk), magnetic tape, or optical media.

Data set naming conventions

• A data set name is 1 to 44 characters long.
• The name is split into qualifiers separated by periods. Maximum 22 qualifiers.
• Each qualifier is 1 to 8 characters: letters A-Z, digits 0-9, national characters (@, #, $), and hyphen (-).
• The first character of a qualifier must be a letter or a national character (@, #, $), never a digit.
• The first qualifier from the left is the high-level qualifier (HLQ). The last one is the low-level qualifier (LLQ).
• Only uppercase characters are used.

Data set organizations

• Sequential (PS): records stored one after another, read from beginning to end.
• Partitioned (PO): a library of members plus a directory. Also called a PDS.
• VSAM: advanced organization for indexed and direct access (KSDS, ESDS, RRDS, LDS).
• Older types (indexed sequential, direct) exist but are rarely used today.

Sequential data sets

• Records are written and read in sequential order, from the first to the last.
• Accessed with the QSAM or BSAM access methods.
• Typical uses: input files, report output, backup files, SYSIN/SYSOUT.
• You cannot read record 100 without reading the 99 before it.

Partitioned data sets (PDS and PDSE)

• A PDS is a library: one or more members plus a directory that lists the members.
• Members are processed individually, for example one COBOL program per member.
• Programmers store source code, JCL, and procedures in PDS libraries.
• A member name is up to 8 characters, starting with a letter or @, #, $.
• PDSE (partitioned data set extended) is the improved version of PDS.
• PDSE advantages: space is reclaimed automatically when a member is deleted, flexible size, can be shared, faster directory searches.

VSAM data sets

• VSAM = Virtual Storage Access Method. It provides more complex functions than other disk access methods.
• KSDS (Key-Sequenced): records are accessed by a key value. Has an index component and a data component.
• ESDS (Entry-Sequenced): records are stored in the order they are added, like a sequential file.
• RRDS (Relative-Record): records are accessed by relative record number.
• LDS (Linear): a byte-stream data set used by DB2 and others.
• VSAM data is stored in control intervals, and records in a KSDS can be read sequentially, randomly by key, or dynamically (both).

Record formats: RECFM, LRECL, BLKSIZE

• RECFM (record format) describes how records are stored. Common values:
• F = fixed length, FB = fixed blocked, V = variable length, VB = variable blocked, U = undefined.
• LRECL (logical record length) is the length of one record in bytes. Example: LRECL=80 for card-image records.
• BLKSIZE (block size) is the size of the physical block written to disk.
• Records are grouped into blocks to save space and I/O time.
• For RECFM=FB, BLKSIZE must be an exact multiple of LRECL. Example: LRECL=80, BLKSIZE=800 means 10 records per block.
• For variable formats, BLKSIZE must be at least LRECL + 4.

Access methods

• An access method is the technique used to store and retrieve data.
• QSAM (Queued Sequential Access Method): for sequential data sets.
• BSAM (Basic Sequential Access Method): for sequential data sets, block by block.
• BPAM (Basic Partitioned Access Method): for reading PDS members.
• BDAM (Basic Direct Access Method): for direct access by address.
• VSAM: for VSAM data sets.

DASD space allocation

• Space is requested in tracks (TRK) or cylinders (CYL): one cylinder = 15 tracks.
• SPACE=(TRK,(primary,secondary,directory)): primary is the first extent, secondary is added if the data set grows, directory is the number of directory blocks for a PDS.
• Space is assigned in extents (contiguous pieces of a volume).
• One DASD volume can hold many data sets, and free space is reused.
• With SMS (System-Managed Storage), a data class can supply the space values automatically.

Ways to allocate a data set

• ISPF data set panel, option 3.2 (allocate new data set).
• TSO ALLOCATE command.
• JCL DD statement with DISP=(NEW,...).
• IDCAMS (Access Method Services) for VSAM.

Define a VSAM KSDS with IDCAMS

• VSAM data sets are created with the IDCAMS DEFINE CLUSTER command.
• KEYS(length offset) gives the key length and its position in the record.
• RECORDSIZE(average maximum) gives the record size.
• The cluster name is what programs use; the DATA and INDEX names are its components.

The VTOC

• The VTOC (Volume Table of Contents) sits on every DASD volume.
• It lists all data sets on the volume and all free space.
• It holds DSCBs (Data Set Control Blocks): labels with each data set's name and location.
• Every volume also has a VOL1 label with the volume serial number (volser).