Setting the file. One moment.
Chapter 54 · Migrating To Amazon Redshift
Subchapter 54.10
references/teradata/reporting.mdMarkdown4 KBView on GitHub
AI-facing knowledge: how to aggregate every phase’s artifacts into a single report.
Roll up the per-phase result/ artifacts into one customer-facing migration report at
output/reporting/result/migration_report.md (plus optional .json for machines / .html for
sharing). The report answers: what was migrated, how much was automated, what passed validation,
how performance compares, and what manual work remains.
| Phase | Artifact | What the report pulls |
|---|---|---|
| Discovery | discovery/result/inventory.json | object counts by type, total tables/rows/size |
| Conversion | converted DDL/SQL/RSQL + conversion/result/manual_review.json | objects converted, confidence scores, manual-review backlog |
| Data migration | data_migration/result/migration_manifest.json | tables loaded, row counts, bytes, per-table status |
| Validation | validation/result/validation_report.json | per-table pass/fail, mismatches |
| Performance | performance/result/perf_baseline.json, perf_compare.json | TD vs RS runtimes, regressions |
Read whatever artifacts exist (a partial run still produces a partial report — note missing phases rather than failing).
manual_review.json.{
"generated_at": "ISO-8601",
"scope": {"databases": 2, "tables": 31, "rows": 19500000},
"summary": {
"objects_total": 0, "objects_converted": 0, "pct_automated": 0.0,
"validation_pass_rate": 0.0, "manual_review_items": 0,
"perf_delta_pct_median": 0.0
},
"phases": {
"discovery": {"status": "complete", "artifact": "discovery/result/inventory.json"},
"conversion": {"status": "complete", "converted": 0, "flagged": 0, "low_confidence": 0},
"data_migration": {"status": "complete", "tables_loaded": 0, "tables_failed": 0, "rows": 0},
"validation": {"status": "complete", "tables_passed": 0, "tables_failed": 0},
"performance": {"status": "complete", "regressions": 0}
},
"manual_review": [
{"object": "db.proc_x", "construct": "QUALIFY+GROUP BY", "confidence": 0.7, "suggestion": "…"}
],
"risks": []
}conversion-rules.md) and manual-review items per object. Count an object as “automated”
when confidence is high and it has zero manual-review items and it isn’t an
unconvertible kind. pct_automated = automated ÷ total objects.manual_review.json) — that’s the actionable
to-do list for the team.manual_review.json schema is defined in conversion-rules.md (producer). The report reads its items[] (object, object_kind, construct, suggested_rewrite, confidence, status) + summary for the backlog and % automated.