Cross Check

Enhanced Definition

A cross-check, in the mainframe context, refers to a verification process where data or calculated results from one source or method are compared against an independent source or an alternative calculation method to ensure accuracy, consistency, and integrity. Its primary purpose is to detect discrepancies, errors, or unauthorized modifications in critical business data or system outputs.

Key Characteristics

- Involves comparing two or more independent sets of data or results that theoretically should match or bear a specific relationship.
- Often implemented within batch processing jobs, typically after data transformation, calculation, or file updates.
- Can utilize various comparison metrics, such as record counts, control totals (e.g., sum of specific fields), hash totals, or detailed field-by-field comparisons.
- Aims to identify data corruption, logical errors in application code, or inconsistencies arising from system failures or manual interventions.
- Critical for maintaining data integrity, supporting auditability, and ensuring the reliability of business-critical applications.

Use Cases

- Financial Reconciliation: Comparing the total debits and credits posted to a general ledger file against the sum of individual transaction records from a daily transaction file using a COBOL program.
- Data Migration Verification: After migrating data from an older VSAM file to a new DB2 table, cross-checking record counts and key field sums between the source and target to ensure no data loss or corruption occurred.
- Report Validation: Verifying the grand total displayed on a generated report (e.g., a monthly customer statement) against an aggregate query directly executed on the underlying DB2 or IMS database.
- Batch Job Integrity: Using a JCL step to run a utility (like SORT or IDCAMS) to count records in an output file and comparing it against an expected count or an input file count from a preceding step.
- System Synchronization: Ensuring consistency between primary and backup data sets or between different copies of a replicated database by periodically comparing key metrics.

Related Concepts

Cross-checking is a fundamental aspect of data quality assurance and error handling within the z/OS environment. It often complements audit trails by providing an independent verification of the data recorded in those trails. It is closely related to reconciliation processes, particularly in financial applications, and is a critical component of robust batch processing workflows, often implemented using COBOL programs, JCL utilities, or REXX scripts. It helps ensure the reliability of data used by CICS transactions or queried by DB2 and IMS applications.

Best Practices:

Automate the Process: Implement cross-checks as automated steps within JCL job streams or application programs to ensure consistent and timely execution.
Define Clear Control Points: Establish specific control totals, record counts, or checksums at various stages of data processing (input, intermediate, output) to facilitate effective comparison.
Log Discrepancies: When a cross-check fails, log detailed information about the discrepancy, including expected vs. actual values, timestamps, and affected data elements, for prompt investigation.
Implement Tolerances: For floating-point arithmetic or complex calculations, define acceptable tolerance levels for minor differences that do not indicate a true error.
Regularly Review Logic: Periodically review and update cross-check logic and control points as business rules, data structures, or application requirements evolve to maintain their effectiveness.