Rapid Recall Key — 01–26
Use only after attempting the prompts.
- Data Integration: movement/consolidation of data within/between stores, applications and organizations.
- Data Interoperability: ability of multiple systems to communicate/exchange through compatible mechanisms.
- Uses include migration/conversion; consolidation/hubs/marts; application/package integration; sharing/distribution; archiving; external ingestion; structured/unstructured integration; operational/analytical support. Any four source-supported examples.
- Principles: enterprise design perspective with iterative delivery; balance local needs with enterprise support/maintenance; business accountability for transformation meaning/change.
- Extract, Transform, Load.
- ELT fits when raw data lands first and the target has strong transformation capability / raw retention is useful.
- Mapping = source-to-target specification of fields/structures, selections, lookups, calculations and transformation rules.
- Latency = time from source generation/change to required target/consumer availability.
- Snapshot = point-in-time set; Delta = only changes since prior transfer.
- CDC identifies inserts/updates/deletes/changed records and transmits deltas instead of full sources.
- High-level data-based techniques: timestamps/change dates; status/change flags/indicators; comparison/list/copy approaches identifying changed rows. Log-based CDC reads DBMS logs.
- Near-real-time permits small delay; synchronous requires current process to wait for immediate downstream response.
- Async = sender continues without waiting for receiver completion/acknowledgment.
- Sync risk = blocking and shared availability dependency.
- Streaming = continuous low-latency event/data processing rather than periodic sets, often in-memory/parallel.
- Replication = same data maintained in several locations for distribution, performance or availability.
- Independent writes at several replicas create conflicts/consistency problems; replication fits controlled exact copies.
- Archive risk = technology/format obsolescence can make retained data unreadable.
- Canonical model = common exchange representation reducing pairwise transformations/semantic inconsistency.
- Point-to-point risk = interface proliferation, transformation inconsistency and rising support/troubleshooting cost.
- Hub advantage = common control/exchange point reduces unique pairwise links and supports common transformation/canonical rules.
- Pub-sub = producer publishes once to topic/service; multiple subscribed consumers receive without separate producer logic.
- Tight = shared availability/immediate dependency; loose = more independent operation, commonly async.
- Orchestration = coordination of execution order, dependencies, timing/frequency, triggers, retry/recovery.
- Process controls include logs, alerts, exception logs, dependency charts and job-clock/timing evidence; any three.
- EAI = applications/modules interact through defined interfaces/APIs rather than direct access to one another’s stores.