A. Incorrect.
Plain CSV lacks transactional guarantees and efficient incremental metadata.

B. Correct.
Delta Lake provides ACID transactions, schema enforcement, and efficient incremental processing for lakehouse workloads.

C. Incorrect.
DNS zones resolve names and do not store streaming data.

D. Incorrect.
A public IP prefix is networking infrastructure, not a data storage layer.

A. Incorrect.
Networking columns do not model dimension history.

B. Incorrect.
Queue and topic names are messaging concepts, not SCD Type 2 history columns.

C. Correct.
Type 2 dimensions commonly track effective start and end dates to preserve historical versions.

D. Incorrect.
DNS fields do not represent dimension validity periods.

A. Incorrect.
Bastion provides secure VM access and does not run data pipelines.

B. Incorrect.
Firewall Manager manages network security policies and does not orchestrate ETL.

C. Incorrect.
Azure DNS hosts DNS zones and does not copy or transform data.

D. Correct.
Azure Data Factory orchestrates data movement and transformation workflows with triggers and activities.

A. Correct.
Row-level security uses predicates to filter rows based on user or execution context.

B. Incorrect.
Cluster autoscaling changes compute size and does not filter rows.

C. Incorrect.
Traffic Manager routes DNS traffic and does not enforce table row filtering.

D. Incorrect.
Blob soft delete protects deleted blobs and does not filter SQL rows.

A. Incorrect.
Forced tunneling is a routing pattern and not a data presentation control.

B. Incorrect.
CDN compression affects web delivery and does not mask SQL data.

C. Correct.
Dynamic data masking can obscure sensitive column values for users without elevated permissions.

D. Incorrect.
Round-robin DNS does not hide database values.

A. Incorrect.
DNS records do not manage pipeline execution.

B. Incorrect.
Access keys alone do not reveal pipeline activity failures.

C. Incorrect.
Front Door WAF logs are unrelated to Data Factory activity runs.

D. Correct.
Data Factory monitoring shows pipeline and activity runs and supports troubleshooting failed activity executions.

A. Incorrect.
A public IP prefix is networking infrastructure and not stream state management.

B. Correct.
Watermarking defines how late streaming events can arrive before state is closed or old events are ignored.

C. Incorrect.
A delete lock prevents resource deletion and does not manage streaming lateness.

D. Incorrect.
An MX record routes email and does not affect event-time processing.

A. Correct.
Replicated distribution can improve joins for small dimension tables by copying them to each compute node.

B. Incorrect.
Hashing on a random GUID can cause unnecessary movement and does not suit all dimensions.

C. Incorrect.
A heap alone does not describe Synapse table distribution.

D. Incorrect.
DNS distribution is not a Synapse table distribution option.

A. Incorrect.
OS info does not provide transaction log file sizing details for the database.

B. Incorrect.
sys.routes relates to Service Broker routing and not data file size.

C. Correct.
sys.database_files exposes file size, max size, and growth information for database files.

D. Incorrect.
External language metadata is unrelated to transaction log growth.

A. Incorrect.
Manual triggers require user action and do not run automatically on file arrival.

B. Incorrect.
Application Gateway listeners route web traffic and do not trigger Data Factory pipelines.

C. Incorrect.
DNS TTL is not a Data Factory trigger type.

D. Correct.
Storage event triggers can start pipelines when blob events occur.