Analyzing File Access Activities

File access activities can often be streamlined to maximize performance. Two of the example application activities called ORDERFL and PRODUCT, probably can be improved. A third activity, called CUSTOMER, seems to be optimally operating. The CICS region-level reports described in this chapter can disclose why transactions spend time in a suspend state as they attempt to access CICS files.

ORDERFL

The ORDERFL access activity processes a VSAM data set by key and accounts for the largest EXCP count among the file access activities listed. The Data Set Characteristics report (See the following figure) shows the EXCP count for file access activities.

Data Set Characteristics Report

The BACK transaction accesses ORDERFL, which appears to contribute to its erratic response time. Both the data component and index component of this data set occupy the same disk storage unit, where usage averages over 30% and approaches 100% for significant periods (as shown in the Time Distribution of Activity Level report, Time Distribution of Activity Level Report). Also, see the I/O Facility Utilization Summary report. Distributing ORDERFL over several disk storage units may significantly reduce average access times but only if the CMXT limitation imposed on BACK is lifted. The CMXT parameter controls the maximum number of transactions of a particular task class that can run simultaneously.

The Transaction Summary report (Transaction Summary Report) shows BACK is a class 3 task. The Classed Tasks report shows that CICS must service class 3 tasks serially, but there were as many as four class 3 tasks awaiting service. Because the maximum allowed is one, BACK must wait each time previous BACK transactions have not been serviced. Enabling concurrent access to several portions of ORDERFL would not affect BACK response time unless you increased the maximum task value.

I/O Facility Utilization Summary Report

PRODUCT

The BACK transaction also accesses PRODUCT, which is another VSAM key-sequenced data set. Although it is being accessed less than 4% of the session time, PRODUCT still could be contributing to the erratic performance of BACK. The Data Set Characteristics report shows excessive EXCPs for the index component of this data set.

Classed Tasks Report

The DASD Usage by Cylinder report (See the following figure) reveals that accesses to PRODUCT concentrate on a single cylinder, which contains the highest-level index control interval. The VSAM default buffer allocation of only one buffer for the index causes VSAM to retrieve all levels of the index from the disk each time PRODUCT accesses a data record. Therefore, increase the number of index buffers to allow more index levels to remain in virtual storage and consequently reduce I/O operations for the indexed component.

DASD Usage by Cylinder Report

CUSTOMER

The Data Set Characteristics report indicates that the CUSTOMER activity also involves access to a VSAM key-sequenced data set and accounts for one-third of all EXCPs issued from within the CICS region.

The Time Distribution of Activity Level report (Time Distribution of Activity Level Report) reveals that this data set is sometimes heavily used. However, it appears from the reports that access attempts to CUSTOMER function very efficiently. The I/O Facility Utilization Summary report shows that the data component of CUSTOMER occupies space on two disk storage units and are about equally accessed; each is never used more than 8% at a time.

Also, both the index component and data component are adequately buffered. VSAM accesses the index component, which is located on a third disk storage unit, infrequently relative to the data component. The data component is well blocked and buffered so that, as you see in the Data Base Operations VSAM Files report (See the following figure), there are nearly 20 times as many operations on this data set as there are physical accesses to the disk storage units.

Data Base Operations VSAM Files Report