IBM Certification Big Data Architect 3 — Questions and Answers
Question 1: In IBM Cloud Pak for Data, what is the primary purpose of the DataStage component?
- Machine learning model training
- ETL and data integration pipelines (Correct answer)
- Real-time event streaming
- Data visualization and dashboarding
Correct answer: ETL and data integration pipelines
IBM DataStage is an ETL tool within Cloud Pak for Data used to design, develop, and run data integration jobs.
Question 2: Which consistency model does Apache Cassandra use, and how does it align with IBM big data architecture recommendations for high availability?
- Strong consistency via two-phase commit
- Eventual consistency with tunable read/write levels (Correct answer)
- Strict linearizability across all replicas
- Snapshot isolation with MVCC
Correct answer: Eventual consistency with tunable read/write levels
Cassandra uses eventual consistency by default but allows tunable consistency levels per operation, enabling architects to balance availability and consistency.
Question 3: When implementing data tiering in an IBM big data solution, which storage tier is most appropriate for infrequently accessed cold data?
- NVMe SSD arrays
- In-memory data grids
- Object storage such as IBM Cloud Object Storage (Correct answer)
- Local SAS disk RAID arrays
Correct answer: Object storage such as IBM Cloud Object Storage
Object storage like IBM Cloud Object Storage is cost-effective and highly durable, making it ideal for cold or archival data tiers.
Question 4: In IBM Db2 Big SQL, what technique allows queries to span both HDFS-resident data and traditional Db2 relational tables in a single statement?
- Federation via database links
- Federated query with external table mappings (Correct answer)
- Replication to a unified staging schema
- ETL pre-processing into a single target
Correct answer: Federated query with external table mappings
Big SQL supports federated queries using external table definitions that map HDFS data sources alongside local Db2 tables.
Question 5: An architect is designing an IBM streaming solution where out-of-order events are common. Which windowing strategy best handles late-arriving data?
- Tumbling windows with fixed offsets
- Sliding windows with overlap
- Session windows with inactivity gaps
- Watermark-based event-time windows (Correct answer)
Correct answer: Watermark-based event-time windows
Watermark-based event-time windowing tracks the progress of event time and allows configurable tolerance for late-arriving data.
Question 6: Which IBM platform component provides automated data quality profiling and cleansing rules as part of a governed data pipeline?
- IBM InfoSphere Information Analyzer (Correct answer)
- IBM MQ Advanced
- IBM Rational Asset Manager
- IBM ILOG CPLEX
Correct answer: IBM InfoSphere Information Analyzer
IBM InfoSphere Information Analyzer profiles data assets to detect quality issues, anomalies, and patterns across enterprise data sources.
Question 7: In a multi-zone IBM Cloud big data deployment, what is the main architectural goal of distributing HDFS DataNodes across availability zones?
- Reducing HDFS block size requirements
- Ensuring rack-aware replication survives zone failure (Correct answer)
- Increasing NameNode read throughput
- Decreasing MapReduce shuffle overhead
Correct answer: Ensuring rack-aware replication survives zone failure
Distributing DataNodes across availability zones ensures that HDFS rack-aware replication policies protect data even if an entire zone fails.
IBM Certification Big Data Architect 3
In IBM Cloud Pak for Data, what is the primary purpose of the DataStage component?