After you define an alternate index cluster and build it with BLDINDEX, applications still need a stable catalog name that means “follow the alternate keys to reach the base records.” That name is the PATH. Naming the path is therefore a user-interface decision as much as a technical one: developers will embed it in COBOL SELECT statements, CICS FCT entries, scheduler metadata, and disaster recovery runbooks. A path name that is short, unique, and obviously not the base cluster name prevents half the midnight mistakes support teams see when someone DELETEs the wrong catalog entry after reading an ambiguous spreadsheet. This page explains how PATH relates to PATHENTRY, compares suffix conventions, and ties naming back to JCL and program configuration.
Think of the base cluster as the warehouse and the AIX as a specialized index card drawer listing coat-check numbers. The PATH is the sign on the door that tells programs which drawer-and-warehouse combination to use together. End users of your API (other developers) should memorize the PATH name, not the internal AIX cluster string, unless your standards explicitly say otherwise. That division of labor mirrors how REST URLs stay stable while internal service names change: here, stability means fewer recompiles when you rebuild an AIX under a new volume strategy but keep the same PATH entry semantics.
DEFINE PATH takes a NAME for the path object and a PATHENTRY referencing the AIX cluster. Typos in either operand create catalog entries that look fine until first OPEN, when VSAM cannot resolve the chain. Read PATHENTRY directly from LISTCAT of the AIX cluster, not from email. If your organization uses multiple alternate indexes on the same base cluster for different keys, each PATH needs a distinct NAME and a PATHENTRY pointing at the correct AIX cluster. Document the matrix: base cluster, key field, AIX cluster, PATH name, owning application.
123DEFINE PATH (NAME(ACCT.CUST.MASTER.DEPT.PATH) - PATHENTRY(ACCT.CUST.MASTER.DEPT.AIX)) /* PATHENTRY must match the AIX cluster name you LISTCAT. */
| Element | Typical pattern | Beginner note |
|---|---|---|
| PATH name | BASE.NAME + .PATH or .ALTKEY.PATH | Signals catalog browsers that this entry is not the physical base cluster. |
| PATHENTRY | Names the AIX cluster created earlier | Must match LISTCAT spelling for the AIX cluster exactly. |
| JCL DSN in jobs | Often the PATH name for alternate-key batch | Keeps application configuration aligned with catalog linkage. |
When a program uses an alternate index, the SELECT often references the PATH name so the runtime resolves keys through the correct catalog object. If developers accidentally point ASSIGN at the base cluster while the program issues START/READ expecting alternate-key behavior, file status codes and runtime errors reflect that mismatch. Naming clarity in meetings (“we open the DEPT path, not the master cluster”) prevents those configuration splits.
Batch jobs that load or reorganize alternate-index structures may reference base cluster, AIX cluster, and PATH names in different steps. Step names in JCL should echo catalog names only when helpful; more importantly, comments in the proc should list all three names whenever any one appears. Future you will not remember that step SORTDEPT depended on PATH ACCT.CUST.MASTER.DEPT.PATH unless it is written down.
CICS FILE entries map datasets used under transactions. Whether CICS uses the base cluster name or a PATH name depends on how file control is configured and whether alternate-key access is required online. When CICS names differ from batch names for the same logical data, incidents arise during parallel testing. Standardize cross-region: if online uses PATH X, document whether batch may still open the base cluster for sequential sweeps or must also use PATH X for consistency.
EXPORT/IMPORT, IDCAMS ALTER, and catalog repair scenarios may touch PATH entries. Because PATH names are part of application configuration, renaming a PATH is often as expensive as renaming a production database connection string. Prefer getting the PATH name right the first time, and if a rename is unavoidable, treat it like an API version bump with coordinated code and JCL changes across systems.
There are two doors into the same playroom. The front door has a big sign that says “Everyone uses this door for recess.” The side door has a sign that says “Teachers only—this door also goes to the playroom but you need the special coat hook map.” The PATH is the side door sign: it still leads to the playroom, but it tells the school which special map to read first. If both signs said the same words, kids would line up at the wrong door every single day.
1. Why do shops avoid making the PATH name identical to the base cluster name?
2. PATHENTRY in DEFINE PATH identifies what?
3. If COBOL SELECT assigns to a PATH name, what happens at open time?