Skip to content

Discrimination, Changed Facts & Readiness

Common distractor traps

  • “Metadata is only column information.” Too narrow; Chapter 12 includes business, technical, operational, lineage, ownership, quality, security, processes, and relationships.
  • “Business Metadata is only used by business people.” Audience does not define the type.
  • “Operational Metadata means operational business data.” It describes processing/access evidence.
  • “Distributed means many repositories plus a central copy.” No: the defining distributed pattern has no persistent central enterprise repository.
  • “Hybrid is centralized with backups.” No: it deliberately combines central persistence and source retrieval.
  • “Bi-Directional means frequent refresh.” No: controlled return flow to sources is required.
  • “Glossary and dictionary are synonyms.” Meaning vs structural detail.
  • “Lineage and impact are synonyms.” Trace path vs consequences of change.
  • “As Designed is more authoritative than code.” It represents intent; As Implemented represents current production behavior.
  • “High completeness proves success.” Coverage can coexist with poor quality, weak security, missing stewardship, and low use.

If the stem says … think first

  • approved definition / owner / rule → Business Metadata
  • table / column / key / schema / mapping → Technical Metadata
  • job log / error / duration / SLA / access pattern → Operational Metadata
  • business terminology → Business Glossary
  • field/data-set characteristics → Data Dictionary
  • where data/systems are located → Directory/Catalog
  • persistent enterprise copy → Centralized
  • query live sources, no persistent store → Distributed
  • selected central + live source detail → Hybrid
  • controlled changes back to source → Bi-Directional
  • System/DataElement/Mapping/Steward entities → Metamodel
  • mapping specification → As Designed
  • actual code path → As Implemented
  • where did it come from/go? → Lineage
  • what breaks if it changes? → Impact Analysis
  • file enters lake → capture Metadata at ingest
  • in-scope vs documented → completeness
  • repository rarely used → usage/adoption
  • conflicting/stale entries → documentation quality

Changed-fact drills

  1. Central copy harvested from tools → Centralized. Remove persistent copy and query sources live → Distributed.
  2. Standards/definitions central, runtime detail source-held → Hybrid. Remove all central persistence → Distributed.
  3. Customer definition → Business Metadata. Change to CUSTOMER_ID VARCHAR(20) → Technical Metadata.
  4. ETL failure timestamp → Operational. Change to source-to-target rule → Technical.
  5. Trace SSN to reports → Lineage. Change to “what breaks if removed?” → Impact Analysis.
  6. Mapping says A→B → As Designed. Actual code A→C→B → As Implemented.
  7. 90% elements documented → strong completeness. Only 5% of users search → usage becomes primary concern.
  8. Catalog exposes protected asset locations broadly → Metadata security problem, even with hidden values.

60-second readiness

Can you explain all ten without notes? 1. B/T/O Metadata. 2. Glossary/Dictionary/Catalog. 3. Four architectures. 4. Five activities. 5. Metamodel. 6. Integrate vs Deliver. 7. Designed vs Implemented. 8. Lineage vs Impact. 9. Why Metadata can be sensitive. 10. Seven metric families.

Final exam order

BUSINESS NEED → STRATEGY → REQUIREMENTS → ARCHITECTURE → METAMODEL/STANDARDS → HARVEST/INTEGRATION → QUALITY/SECURITY/GOVERNANCE → DELIVERY/USE → LINEAGE/IMPACT → METRICS/IMPROVEMENT

← Tier 1 · Visual Atlas →