NCL File Processing › Work with UDBs › VSAM Considerations for Alternate Indexes

VSAM Considerations for Alternate Indexes

An alternate index is an individual VSAM data set, maintained by VSAM, which contains information and keys to the base cluster VSAM data set to which the index applies. It can be regarded as providing the user with a different view of the data within the base cluster. As changes are made to records on the base cluster, VSAM (provided that the correct options are defined) automatically updates the alternate indexes associated with the cluster.

VSAM relates an alternate index to its target base cluster with a definition known as a path. It is this path which NCL uses to retrieve data using keys from the alternate index.

A given base cluster can have many alternate indices, depending on the complexity of the record formats maintained on the UDB and the number of different ways those records are to be accessed.

The process to establish a new base cluster with an alternate index is:

• Define the base cluster using the IDCAMS DEFINE CLUSTER function.
• Load the base cluster. This is done using the VSAM IDCAMS utility REPRO statement.
• Define the alternate index using the IDCAMS DEFINE AIX function.
• Define the path using the IDCAMS DEFINE PATH function.
• Build the alternate index using the IDCAMS BLDINDEX function.

  This cannot be done for an empty base cluster.

  Add DD statements to the system JCL for both the base cluster and the path (but not the alternate index).

• Assign file IDs to UDBs using the UDBCTL statement.

Notes:

• When defining a base cluster with an alternate index, both the base cluster and the alternate index must be defined with VSAM SHAREOPTIONS
• of (3 3). This is necessary as two ACBs will be used, one to reference the base cluster and a second to reference the path. A reference to a path causes both the base cluster and its associated alternate index to be opened. VSAM therefore sees two concurrent users of the base cluster.
• If duplicate keys are to be allowed on the alternate index, the NONUNIQUEKEY operand must be used when the index is defined. Duplicate keys on the alternate index are likely when the alternate index key is positioned over part of the base key of a cluster.
• To establish an alternate index for an existing base cluster, the same procedure is followed but with the omission of the first two steps (that is, defining and loading the base cluster).