Testing Data Products, Replication Flows and Analytics Models in SAP Datasphere and Business Data Cloud

Keep this: job aid

Data-layer testing and sign-off card

• Map every hand-off: replication, installation, transformation, model, consumption; reconcile each.
• Fix the window: closed period for regression; last delta timestamp as cut-off for open periods.
• Agree zero tolerance for counts and stored amounts before testing; rounding tolerance only for translation.
• Replication: test initial load by dimension, then known inserts, updates, deletes; stop-and-resume without gaps.
• Watch: Data Integration Monitor (Flows); source DHCDCMON and SLG1; SAP Notes 2890171 and 3297105.
• Reconcile counts → totals → key-level diffs; store each check as a pair of identical-selection queries.
• Models: rule checks on known cases (90001877: 12,000.00 / 7,500.00 / 4,500.00), plant 1720 at 156.00, credit memos, exclusions.
• Regression baseline: counts, totals, structure, versions, rule checks; rerun on SAP updates, releases, support packages.
• Classify every difference: expected (with source) or defect.
• Evidence pack: 9 sections, rerunnable, untested scope stated; financial differences, access failures and delta loss always block.