Scenario Lab 07–12
7 — Vendor system exposes change table
Stem: application stores inserts, updates and deletes in a dedicated log table.
- Primary: source-supported CDC.
- Supporting: Integration, audit/control, operations.
- Roles: source owner, integration engineer, DW operations.
- Best: log-table delta CDC using explicit change records.
- Weaker: full compare despite good change evidence; unnecessary processing.
- Changed: no usable change indicator → full comparison may become necessary.
- Source: pp. 371–374.
8 — Five-minute mini-batches
Stem: loads fire every five minutes or after a source threshold.
- Primary: low-latency source-side accumulation.
- Best: trickle feed.
- Weaker: call it streaming solely because frequent; streaming clue is target queue/buffer.
- Changed: tiny events published to shared bus → messaging.
- Source: pp. 372–374.
9 — Events on enterprise bus
Stem: targets subscribe independently to transaction events.
- Primary: event-based middleware integration.
- Roles: integration architect, message-platform team, source/target owners.
- Best: messaging through bus/pub-sub.
- Weaker: trickle; no source-side mini-batch clue.
- Changed: continuous direct arrivals accumulate in target queue → streaming.
- Source: pp. 372–374.
10 — Target queue
Stem: warehouse continuously receives events and processes them in order from its buffer.
- Primary: target-side continuous-flow accumulation.
- Best: streaming.
- Weaker: messaging when no middleware bus is the accumulation/distribution point.
- Changed: source accumulates count/time threshold before sending → trickle feed.
- Source: pp. 372–374.
11 — Dashboard request starts the project
Stem: team immediately catalogs tables without clarifying goals/questions.
- Primary: requirements sequence/business alignment.
- Supporting: Architecture, BI portfolio, source assessment, Governance.
- Roles: sponsor, analysts, DW/BI product team, architect.
- Best: clarify goals, decisions, KPIs, users, grain/history/latency/security before committing to sources/tools.
- Weaker: build from easiest tables; lets source availability drive business scope.
- Changed: questions/KPIs already clear → profiling/mapping can proceed.
- Source: pp. 374–376.
12 — Equivalent fields, different names
Stem: three billing systems label customer identifiers differently.
- Primary: source-to-target semantic mapping.
- Supporting: Metadata, Modeling, Integration, Governance.
- Roles: Modeler, integration engineer, Steward/SME.
- Best: source-to-target mapping using common logical taxonomy/enterprise semantics.
- Weaker: automatically label different names as DQ defects; they may be valid source-specific semantics.
- Changed: malformed/invalid values → remediation becomes primary.
- Source: pp. 375–379.