Problem indicators in an optimization report

2.1+

This section describes the different analysis states of an optimization report, and focuses on reasons why an analysis may have failed.

IBM® watsonx Code Assistant for Z Code Optimization Advice can be used to analyze one or many programs at once. Each individual program in an application included in the analysis will have a complete analysis or failed analysis state.

A program will either have a complete analysis or a failed analysis.

A complete analysis is generated when all input data required to analyze the program are valid. Input data includes:
  • Program
    Note: 2.4+

    Only Enterprise COBOL V4 and V6 programs will have a complete analysis. Analysis performed on V5 or pre-V4 versions of the COBOL compiler will result in a failed analysis.

  • ADATA: Used to detect problems, determine data item names, and gain an understanding of the COBOL source code
  • Listing file: Used to detect problems, determine line numbers in compiled code, and connect the ADATA and profile
  • Profile: Used to determine performance impact of problems
Note: There are cases where all expected input data is provided and valid, but an error may occur where data items could not be determined, or the performance samples could not be measured, resulting in an unknown status classification.

Analysis is skipped and will be reported as failed if any of the required input data is missing, or if any of the inputs are invalid. In some cases a missing or invalid profile input may prevent a report from being generated at all. Hover over the failed analysis icon or select the failed program to learn why the error occurred.

Priority indicators

Z Code Optimization Advice ranks all problems that were detected and prioritizes them based on the impact that each problem has on performance. Z Code Optimization Advice uses a combination of runtime profiling data, source code analysis, and executable binary analysis to determine this metric.
Note: Z Code Optimization Advice uses a best-effort ranking algorithm that combines IBM's expertise in optimizing COBOL programs with the performance information gathered by profiling the running application. It is not intended to be prescriptive. The best way to accurately determine the performance impact of a problem is to reanalyze a program after the problem has been resolved and compare the new CPU usage value with the original. For information on comparing reports, refer to Comparing optimization reports.
Priorities:
  • Critical: If addressed, this problem is expected to have a highly significant impact on relative performance
  • High: If addressed, this problem is expected to have a significant impact on relative performance
  • Medium: If addressed, this problem is expected to have moderate impact on relative performance
  • Low: If addressed, this problem is expected to have small impact on relative performance
  • Unmeasured: There are two main reasons for why a problem is unmeasured:
    • The performance impact of this problem could not be determined. Problems in this category may have an impact that ranges from highly significant to practically none at all
    • The performance of the problem could not be measured because zero samples were detected