The virtual resource pool for all components of the clusters or alternate indexes must be successfully built before any open is issued to use the resource pool; otherwise, the results might be unpredictable or performance problems might occur. To specify the BUFFERS, KEYLEN, and STRNO parameters of the BLDVRP macro, you must know the size of the control intervals, data records (if spanned), and key fields in the components that will use the resource pool. You must also know how the components are processed. You can use the SHOWCAT and SHOWCB macros, or the access method services LISTCAT command to get this information.

For example, to find the control interval size using SHOWCB: open the data set for nonshared resources processing, issue SHOWCB, close the ACB, issue BLDVRP, open the ACB for LSR or GSR.

Tip: Because Hiperspace buffers are in expanded storage, you do not need to consider their size and number when you calculate the size of the virtual resource pool.

For each VSAM cluster that will share the virtual resource pool you are building, follow this procedure:

1. Determine the number of concurrent requests you expect to process. The number of concurrent requests represents STRNO for the cluster.
2. Specify BUFFERS=(SIZE(STRNO+1)) for the data component of the cluster.
   • If the cluster is a key-sequenced cluster and the index CISZ (control interval size) is the same as the data CISZ, change the specification to BUFFERS=(SIZE(2 X STRNO)+1).
   • If the index CISZ is not the same as the data component CISZ, specify BUFFERS=(dataCISZ(STRNO+1),indexCISZ(STRNO)).

Following this procedure provides the minimum number of buffers needed to support concurrently active STRNO strings. An additional string is not dynamically added to a shared resource pool. The calculation can be repeated for each cluster which will share the resource pool, including associated alternate index clusters and clusters in the associated alternate index upgrade sets.

For each cluster component having a different CISZ, add another ‘,SIZE(NUMBER)’ range to the ‘BUFFERS=’ specification. Note that the data component and index component buffers may be created as one set of buffers, or, by use of the ‘TYPE=’ statement, may be created in separate index and data buffer sets.

Additional buffers may be added to enhance performance of applications requiring read access to data sets by reducing I/O requirements. You should also consider the need for cross-region or cross-system sharing of the data sets where modified data buffers must be written frequently to enhance read and update integrity. Many buffers is not usually an advantage in such environments. In some applications where a resource pool is shared by multiple data sets and not all data set strings are active concurrently, less than the recommended number of buffers may produce satisfactory results.

If the specified number of buffers is not adequate, VSAM will return a logical error indicating the out-of-buffer condition.