Data Warehousing on AWS Training Data Warehousing on AWS: Architecture Design 3 — Questions and Answers
Question 1: A data architect wants to separate storage costs from compute costs in a modern cloud data warehouse. Which Redshift deployment mode best achieves this?
- Redshift RA3 nodes with managed storage (Correct answer)
- Redshift DC2 nodes with local SSD storage
- Redshift DS2 nodes with HDD storage
- Redshift Serverless with on-demand capacity
Correct answer: Redshift RA3 nodes with managed storage
RA3 nodes decouple compute and storage by using high-performance SSDs as cache while offloading data to S3-backed managed storage.
Question 2: What is the recommended approach for sharing live data across multiple Redshift clusters without copying or moving data?
- Cross-region snapshots
- Redshift Data Sharing (Correct answer)
- S3 replication with Spectrum
- AWS Glue ETL jobs
Correct answer: Redshift Data Sharing
Redshift Data Sharing allows producer clusters to share live data with consumer clusters without data movement or copying.
Question 3: In the AWS lake house architecture, which service serves as the centralized metadata catalog connecting Redshift, Athena, and EMR?
- Amazon CloudWatch
- AWS Glue Data Catalog (Correct answer)
- Amazon DynamoDB
- AWS Lake Formation
Correct answer: AWS Glue Data Catalog
The AWS Glue Data Catalog provides a unified metadata repository that Athena, Redshift Spectrum, and EMR can all use to query S3 data.
Question 4: Which Redshift sort key type is most effective when queries frequently filter on a single column using range predicates?
- Interleaved sort key
- Compound sort key (Correct answer)
- AUTO sort key
- No sort key
Correct answer: Compound sort key
Compound sort keys work best when queries filter predominantly on the leading sort column, enabling zone map pruning for range predicates.
Question 5: A company is designing a data warehouse and needs to preserve a complete history of changes to dimension attributes over time. Which technique should they use?
- Type 1 SCD (overwrite)
- Type 2 SCD (add new row) (Correct answer)
- Type 3 SCD (add new column)
- Type 0 SCD (retain original)
Correct answer: Type 2 SCD (add new row)
Type 2 SCD preserves full history by inserting a new row for each change, with effective date ranges or current-row flags.
Question 6: Which AWS service is typically used to orchestrate complex, multi-step ETL pipelines that load data into a Redshift data warehouse?
- Amazon SQS
- AWS Step Functions (Correct answer)
- Amazon Kinesis Firehose
- Amazon SNS
Correct answer: AWS Step Functions
AWS Step Functions orchestrates multi-step workflows, coordinating Lambda, Glue, and other services in ETL pipelines.
Question 7: In Amazon Redshift, what is a 'slice' and why does it matter for architecture design?
- A logical partition of data within a node that enables parallel processing (Correct answer)
- A backup segment stored in S3 for disaster recovery
- A query queue used by WLM to prioritize workloads
- A read replica that serves BI tool connections
Correct answer: A logical partition of data within a node that enables parallel processing
Each compute node is divided into slices, each with its own CPU and memory, enabling data to be processed in parallel across all slices.
Data Warehousing on AWS Training Data Warehousing on AWS: Architecture Design 3
A data architect wants to separate storage costs from compute costs in a modern cloud data warehouse.
Which Redshift deployment mode best achieves this?