Rapid Recall Key 01–38
Give full credit only when you can state the deciding distinction—not merely a familiar keyword.
- DW/BI management: design, build, populate, deliver, govern, operate and improve integrated analytical data/products for reporting, analysis, compliance and decisions.
- Goals: maintain an integrated analytical environment; make reliable analytical capabilities available to generate insight/better decisions.
- Principles: business goals; end in mind; design globally/build locally; summarize/optimize last; transparency/self-service; Metadata with warehouse; cross-discipline collaboration; one size does not fit all.
- BI / DW / Warehousing: analysis + enabling tech / integrated historical environment / processes that collect-cleanse-transform-integrate-load-maintain it.
- EDW: enterprise-wide because subjects/sources are integrated/standardized for consistent enterprise analysis, not because of the label.
- Inmon four: subject-oriented, integrated, time-variant, non-volatile.
- Integrated ≠ copied: reconcile keys, codes, structures, definitions and semantics rather than preserve source inconsistency.
- Non-volatility: preserves historical evidence/reproducibility instead of overwriting prior states.
- Atomic vs summary: lowest useful event detail/flexibility vs aggregates for specific usability/performance.
- CIF: applications/sources, staging, integration/transformation, ODS as needed, enterprise DW, marts/OpDM, BI/access, Master/Reference/external context.
- ODS vs DW: current/near-current, lower latency, more volatile, shorter history vs integrated historical, stable/non-volatile.
- OpDM vs mart: tactical current/near-term presentation often sourced from ODS vs classic historical analytical subset.
- Master/Reference support: consistent entity identity, codes, classifications, hierarchies/shared context; often supports conformed dimensions.
- Kimball: business-process dimensional facts/dimensions integrated through conformed dimensions/facts and DW bus.
- Fact vs dimension: quantitative process measure vs descriptive context.
- Conformed dimension: defined/used consistently across multiple fact areas/marts for comparison/integration.
- Bus matrix: business processes/fact areas crossed with dimensions; scopes conformance/increments.
- Inmon vs Kimball: central normalized enterprise integration vs dimensional process integration through conformance/bus.
- Architecture components: sources, integration, staging, ODS, historical DW, marts/cubes/presentation, BI/access plus Metadata/DQ/Security/Governance/operations.
- Staging vs central DW: intermediate preparation/no normal user analysis vs persistent integrated historical analytical store.
- ODS vs mart: current integrated operational use vs targeted analytical subset/presentation.
- OLAP families: ROLAP, MOLAP, HOLAP.
- Historical load vs ongoing: initial/backfill vs recurring refresh/deltas after baseline.
- History from requirements: current state, trend, point-in-time, audit and predictive needs imply different grain/retention/change strategies.
- Data Vault: granular historical relationships/keys, traceability and ability to rebuild/adapt downstream presentation.
- Five CDC: timestamp; log table; transaction log; message delta; full comparison.
- Timestamp vs log table: timestamp selects by high-water mark but deletes not inherent; log table holds explicit change records including deletes if designed.
- Transaction log vs message: DBMS transaction evidence, detailed/delete-aware/platform dependent vs application-published change events through messaging.
- Full comparison: when no reliable change indicator/log/event exists or full-state comparison is acceptable despite cost.
- Low latency: trickle, messaging, streaming.
- Accumulation: trickle=source; messaging=bus; streaming=target queue/buffer.
- Isolate volatility: preserve durable historical DW role while satisfying current/low-latency needs separately.
- Six activities: Requirements → Define/Maintain Architecture → Develop DW/Marts → Populate → Implement BI Portfolio → Maintain Data Products.
- DW vs operational requirements: exploratory/integrated questions, grain/history/DQ/lineage/latency/security/performance/user communities vs executing defined transactions/workflows.
- Interview capture: goals/decisions, users, KPIs/questions, source/data, grain/history, latency, DQ, security/access, performance/availability, Metadata/lineage, support/adoption.
- Architecture: source/destination/timing/rationale/movement, staging/integration/storage/history/presentation plus Metadata, security, performance, availability/scalability, support/release evolution.
- Three tracks: Data; Technology; BI/Delivery tools.
- Source-to-target mapping: documents source element → target element and rules/logic used to populate target; core lineage Metadata.
Source range: pp. 361–379.