Mainframe documentation loves acronyms. VSAM conversations pile KSDS, CI, CA, RBA, AIX, PATH, and IDCAMS into single sentences, which is efficient for experts and opaque for newcomers. This page is a structured glossary: short definitions you can skim before a stand-up, plus enough context to know which deep-dive tutorial to open next. It is not a substitute for IBM manuals—those remain authoritative for edge parameters—but it is a Rosetta stone for reading LISTCAT printouts, JCL errors, and senior engineers’ whiteboard sketches without stopping the meeting every ninety seconds.
Scan the tables first. When a term sparks confusion, open the linked tutorial from the related concepts or related pages sections at the bottom. In real incidents, combine glossary recall with LISTCAT evidence: terms mean little if the catalog says your object is LINEAR when you thought it was KSDS. Treat words and catalog facts as paired sources of truth.
| Term | Meaning |
|---|---|
| KSDS | Key-sequenced dataset: records ordered by a primary key; supports random, sequential, and dynamic access; has separate index and data components. |
| ESDS | Entry-sequenced dataset: records in load order; byte-level (RBA) addressing; no native primary key like KSDS; deletes are not physical in the classic sense. |
| RRDS | Relative record dataset: fixed slots addressed by relative record number; good for direct slot access when the application model is array-like. |
| LDS | Linear data set: byte stream without VSAM record structure as used by KSDS; underpinning technology for Db2 and other subsystems. |
| AIX | Alternate index: secondary keyed structure over a KSDS or ESDS base cluster; built and maintained with DEFINE AIX/BLDINDEX patterns. |
| PATH | Catalog object connecting an AIX to a base cluster so applications can open one stable name for alternate-key access. |
| Term | Meaning |
|---|---|
| Cluster | Logical VSAM file umbrella in the catalog, implemented by one or more physical components depending on organization. |
| Data component | Physical dataset holding user records (and CI structure) for the cluster. |
| Index component | KSDS structure storing keys and pointers for keyed lookup; sequence set and index set live here conceptually. |
| CI | Control interval: the unit of transfer and space management inside VSAM; tuning CI size affects splits and I/O size. |
| CA | Control area: contiguous group of CIs; CA splits are more expensive than CI splits and signal space design stress. |
| FREESPACE | Percentage parameters reserving empty space inside CIs and CAs to absorb inserts without immediate splits. |
| High-used RBA | Marker for how far into the data component meaningful bytes extend for ESDS-style growth thinking. |
| Term | Meaning |
|---|---|
| RBA | Relative byte address from dataset origin; used heavily in ESDS and in some low-level diagnostics. |
| RRN | Relative record number slot index for RRDS addressing. |
| Key | Field within a KSDS record used for ordering and direct reads; defined with KEYS(length,offset) on DEFINE CLUSTER. |
| Term | Meaning |
|---|---|
| DEFINE CLUSTER | IDCAMS command to create a VSAM cluster and its components with attributes like RECORDSIZE, KEYS, and space. |
| DELETE | IDCAMS command to remove catalog entries and optionally uncatalog or scratch underlying objects per operands. |
| LISTCAT | IDCAMS reporting command to display catalog entries; the first troubleshooting tool for names and attributes. |
| REPRO | IDCAMS command to copy or convert datasets, often used for VSAM loads and backups. |
| EXPORT / IMPORT | Portable cluster packaging and restore flow for movement between systems or disaster recovery when used with supporting procedures. |
| VERIFY | IDCAMS command to reconcile end-of-data and catalog flags after abnormal close scenarios, per documented conditions. |
A sphere is the conceptual grouping of a base cluster and its dependent alternate indexes and paths. People say “in the sphere” to remind listeners that DELETE or EXPORT decisions may ripple beyond the base cluster name printed largest on the diagram.
SHAREOPTIONS govern cross-system and cross-region sharing behavior at the VSAM level. They interact with ENQ, RESERVE, and DFSMS capabilities. Beginners should memorize that SHAREOPTIONS are not a substitute for application serialization logic—they describe what VSAM allows, not what your program correctly coordinates.
Record Level Sharing introduces locking semantics for shared VSAM access in CICS and other environments. RLS discussions borrow terms from distributed databases; under the hood, VSAM and coupling facility structures still matter. See the RLS architecture page when you graduate from batch-only thinking.
Storage administrators talk about STORCLAS, DATACLAS, and MGMTCLAS as the SMS triplet. None of these replace DEFINE CLUSTER, but they influence where new allocations land, which technical defaults apply, and how backup or migration policies treat the dataset over months. When someone says “that VSAM picked up the wrong DATACLAS,” they mean ACS assigned technical defaults that conflict with your record length or CI expectations—not that VSAM misread your application logic. Learning SMS acronyms alongside VSAM acronyms prevents blame storms between application teams and infrastructure teams during space failures.
CI size tuning conversations mention splits, free space, and CA size in one breath. CI split means a single control interval fractured because an insert or lengthened record no longer fit; VSAM redistributes records and adjusts the index. CA split means an entire control area boundary was stressed, which is heavier work. Buffer tuning discussions mention BUFND and BUFNI (data and index buffers), STRNO (strings for concurrent requests), and sometimes ACCBIAS for access bias hints. You do not need to memorize every parameter on day one, but you should recognize the words as levers that exist so you can open the dedicated tuning pages instead of guessing from folklore.
FILE STATUS returns two-byte codes after VSAM operations in COBOL programs. CICS file control verbs (READ, WRITE, STARTBR, READNEXT) also surface response codes and conditions. The glossary words cluster, PATH, and AIX reappear in SELECT statements, FCT entries, and transaction dumps. When you debug, write the chain on paper: program name, DD or file definition, dataset or PATH name, cluster organization, and the last IDCAMS attribute change. That five-tuple removes ambiguity faster than rereading the same glossary row twelve times.
A glossary is like the picture dictionary on the classroom shelf. When someone says “CI,” you do not have to guess whether they mean cat ice cream; you flip to the picture that shows a LEGO plate of a certain size fitting into a bigger LEGO board. VSAM’s plate is the control interval. When someone says “KSDS,” the picture shows toys sorted by number tags so you can jump straight to toy number 42. Different pictures, same shelf.
1. Which dataset type uses a primary key for ordering and keyed reads?
2. What does CI stand for?
3. Which utility command is the usual first step to see catalog attributes for a cluster?