100% Latest & Updated Amazon AWS Certified Data Engineer - Associate DEA-C01 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 770
A data engineering service uses one unrestricted role for production ETL, development notebooks and reporting. Which engineering change best addresses the underlying issue?
- Apply approved tokenization, masking, or projection before data exits the trusted boundary.
- Scope KMS actions to intended keys and add appropriate conditions where supported.
- Apply supported dynamic data masking or carefully governed masked views and role access.
- Separate service roles and policies by environment, workload and required resources.
Explanation
Correct Answer
D:Separate service roles and policies by environment, workload and required resources. It directly addresses the case cause (Shared broadly privileged identities obscure duties and violate least privilege.) and should be verified by: Development cannot write production data while ETL still runs correctly.
Incorrect Answers
A:That change addresses a DynamoDB-backed application needs to redact sensitive values before exporting analytics but relies only on table encryption; it would not repair this incident.
B:That change addresses a Glue data job decrypts source objects using a wildcard KMS permission spanning unrelated client keys; it would not repair this incident.
C:That change addresses a Redshift view exposes unmasked customer SSNs to a general BI role despite a requirement for partial disclosure; it would not repair this incident.
Question 771
After resolving the following incident, which ongoing safeguard most directly prevents recurrence? A SageMaker Catalog project shares a data product broadly across teams without recording owner and allowed purpose.
- Link pipeline run IDs to audit role and request context.
- Audit Redshift database roles and grants periodically.
- Validate privacy controls beyond simply checking that raw strings changed.
- Require owner and data-classification fields during data-product publication.
Explanation
Correct Answer
D:Require owner and data-classification fields during data-product publication. This safeguards against a recurring cause: Business catalog sharing lacks stewardship and usage-governance metadata.
Incorrect Answers
A:That safeguard addresses another case: An ETL role invokes KMS Decrypt repeatedly, but the pipeline's incident review examines only application logs without recording AWS API actor identity. It is not the control for this recurring failure.
B:That safeguard addresses another case: A Redshift data analyst has a role granting broad schema CREATE and DROP despite requiring only SELECT on a few views. It is not the control for this recurring failure.
C:That safeguard addresses another case: A tokenization pipeline uses a deterministic unkeyed hash on low-entropy postal identifiers and calls them irreversibly anonymous. It is not the control for this recurring failure.
Question 772
A data quality gate compares raw and target record totals without accounting for intentional filter exclusions, falsely failing clean batches. Select THREE complementary measures that correctly repair and validate, and prevent recurrence of this failure.
- Reconcile accepted, filtered and quarantined records with explicit balance equations.
- A product ID such as 00027 remains unchanged after the approved cleaning step.
- A known filtered fixture passes with every source record accounted for.
- Maintain row-count ledgers through ETL stages.
- Configure supported persistent log delivery to a controlled durable location.
- Disable all reconciliation because filtering changes totals.
Explanation
Correct Answers
A:Reconcile accepted, filtered and quarantined records with explicit balance equations. The underlying cause is: The reconciliation rule ignores documented transformation semantics and accepted rejects.
C:A known filtered fixture passes with every source record accounted for. This is the observable result required before accepting the correction.
D:Maintain row-count ledgers through ETL stages. This protects against the following risk: The reconciliation rule ignores documented transformation semantics and accepted rejects.
Incorrect Answers
B:This would validate a different case: An analyst creates a Glue DataBrew visual cleaning step that trims all leading zeros from product IDs, breaking lookups. It does not verify the outcome required here.
E:That corrective action is for a different case: An EMR job intermittently fails due to capacity interruptions but its logs are stored only on the terminated instance. It does not resolve this incident.
F:Disable all reconciliation because filtering changes totals. That shortcut conflicts with the stated requirement.
Question 773
A Step Functions Standard workflow waits for an external vendor callback but the task token is lost in logs. Which technical dependency most directly explains this behavior?
- Processing exceeds configured consumer poll or session expectations without adequate checkpoint strategy.
- Transformation discards intended field types and semantic metadata.
- Network routing or security groups block the ETL job's actual source path.
- The callback correlation contract is missing or insecure.
Explanation
Correct Answer
D:The callback correlation contract is missing or insecure. The corresponding fix is to persist scoped task-token correlation securely and use supported callback completion.
Incorrect Answers
A:This diagnosis addresses another AWS case: An MSK consumer group repeatedly rebalances during long processing, causing duplicate deliveries. The dependency here is different.
B:This diagnosis addresses another AWS case: A clickstream pipeline converts all nested attributes to strings before storage, making numeric analytics unreliable. The dependency here is different.
C:This diagnosis addresses another AWS case: A Glue connection to a source RDS database times out, although the JDBC credentials are valid. The dependency here is different.
Question 774
After resolving the following incident, which ongoing safeguard most directly prevents recurrence? An engineer adds a new processing step to an AWS Glue workflow but never connects its trigger to the existing dependency chain.
- Track cardinality and group-level drift in quality dashboards.
- Test workflow reachability and trigger states after DAG edits.
- Monitor per-task skew rather than average cluster CPU alone.
- Document allowed lateness and watermark monitoring by source.
Explanation
Correct Answer
B:Test workflow reachability and trigger states after DAG edits. This safeguards against a recurring cause: The new job is present but unreachable from the workflow's actual trigger graph.
Incorrect Answers
A:That safeguard addresses another case: A Spark aggregation produces incorrect category totals because one hot key is dropped after a task retry and the target doesn't reconcile. It is not the control for this recurring failure.
C:That safeguard addresses another case: An ETL task in Glue times out only when a small fraction of keys dominate the daily partition. It is not the control for this recurring failure.
D:That safeguard addresses another case: A nightly data product uses a moving seven-day window but incoming source events arrive late and are excluded without a watermark policy. It is not the control for this recurring failure.
Question 775
A Kinesis consumer reports healthy CPU yet its IteratorAge grows for hours during normal traffic. Select TWO complementary measures that correctly repair and validate this failure.
- Use CPU below 50% as proof there is no streaming lag.
- Athena reads only successfully committed transform output after the Glue run ends.
- Select only needed fields and apply effective partition predicates where applicable.
- Consumer lag decreases under a measured production-like load.
- Examine iterator age, shard utilization, downstream latency and consumer parallelism.
Explanation
Correct Answers
D:Consumer lag decreases under a measured production-like load. This is the observable result required before accepting the correction.
E:Examine iterator age, shard utilization, downstream latency and consumer parallelism. The underlying cause is: Consumption lag is growing despite adequate compute metrics and may indicate insufficient read/processing throughput.
Incorrect Answers
A:Use CPU below 50% as proof there is no streaming lag. That shortcut conflicts with the stated requirement.
B:This would validate a different case: A Step Functions data pipeline invokes Glue using a request/response integration and immediately launches Athena against unfinished output. It does not verify the outcome required here.
C:That corrective action is for a different case: An Athena query uses SELECT * on a wide Parquet table, although the report needs two measures and a date. It does not resolve this incident.
Question 776
A team applies a reviewed fix for this situation: A large historical data export shares the same small API rate limit as near-real-time orders and delays critical events. Which observed outcome would most convincingly validate the fix?
- Critical orders stay within latency SLA during a backfill rehearsal.
- A fixture with two nested lines produces correct line and order totals.
- A replayed historical event receives its original applicable classification.
- Only permitted attributes enter the model request and classification remains usable.
Explanation
Correct Answer
A:Critical orders stay within latency SLA during a backfill rehearsal. This verifies the result of separate ingestion capacity budgets and backpressure with appropriate queues or schedules.
Incorrect Answers
B:This would verify another situation: A Glue job reading nested JSON flattens arrays incorrectly and duplicates parent order totals. It does not validate the current case.
C:This would verify another situation: A Glue transform joins events to reference data refreshed weekly, but old events are silently reclassified on every rerun. It does not validate the current case.
D:This would verify another situation: An ETL pipeline calls a Bedrock model on full records containing account secrets merely to extract public product categories. It does not validate the current case.
Question 777
A SageMaker Catalog project shares a data product broadly across teams without recording owner and allowed purpose. Which technical dependency most directly explains this behavior?
- Metadata discovery does not imply Lake Formation table data permission.
- The role trust policy does not admit the actual service principal.
- Business catalog sharing lacks stewardship and usage-governance metadata.
- Cryptographic permissions exceed the data processing task's required keys.
Explanation
Correct Answer
C:Business catalog sharing lacks stewardship and usage-governance metadata. The corresponding fix is to publish ownership, purpose, classification and scoped project access before sharing.
Incorrect Answers
A:This diagnosis addresses another AWS case: A Lake Formation-enabled S3 table is visible in Glue Data Catalog, but Athena SELECT is denied to the analyst role. The dependency here is different.
B:This diagnosis addresses another AWS case: A Step Functions state invokes a Lambda using a role that trusts EC2 instead of the Step Functions service. The dependency here is different.
D:This diagnosis addresses another AWS case: A Glue data job decrypts source objects using a wildcard KMS permission spanning unrelated client keys. The dependency here is different.
Question 778
A Step Functions data pipeline invokes Glue using a request/response integration and immediately launches Athena against unfinished output. Which engineering change best addresses the underlying issue?
- Aggregate by validated stable user identity and the intended time-window rules.
- Use an ordered and explicitly framed rolling calculation on normalized dates.
- Set an appropriate retention and controlled access policy for log groups.
- Use supported wait-for-completion integration or a correctly bounded job-status polling loop.
Explanation
Correct Answer
D:Use supported wait-for-completion integration or a correctly bounded job-status polling loop. It directly addresses the case cause (The orchestration integration returns before the asynchronous ETL work finishes.) and should be verified by: Athena reads only successfully committed transform output after the Glue run ends.
Incorrect Answers
A:That change addresses a marketing report uses COUNT(*) of event records as its active user metric and overstates adoption when users retry; it would not repair this incident.
B:That change addresses a data science notebook uses a rolling 7-day average without ordering by event date, so results change with input row order; it would not repair this incident.
C:That change addresses a Lambda processor logs errors but its CloudWatch Logs retention is shorter than the postmortem evidence requirement; it would not repair this incident.
Question 779
An Athena automated report tries to read query results immediately after StartQueryExecution returns a query ID. Which technical dependency most directly explains this behavior?
- The query unnecessarily reads columns beyond its actual projection.
- Events and distinct users are different measurement grains.
- The report does not wait for the asynchronous query to reach SUCCEEDED.
- The sample is too small or unrepresentative for a claimed whole-population guarantee.
Explanation
Correct Answer
C:The report does not wait for the asynchronous query to reach SUCCEEDED. The corresponding fix is to poll GetQueryExecution with bounded backoff and handle FAILED/CANCELLED before reading results.
Incorrect Answers
A:This diagnosis addresses another AWS case: An Athena query uses SELECT * on a wide Parquet table, although the report needs two measures and a date. The dependency here is different.
B:This diagnosis addresses another AWS case: A marketing report uses COUNT(*) of event records as its active user metric and overstates adoption when users retry. The dependency here is different.
D:This diagnosis addresses another AWS case: A streaming pipeline evaluates quality from a random sample of 10 messages and assumes it proves a million-row batch is correct. The dependency here is different.
Question 780
A team applies a reviewed fix for this situation: An AWS Config rule flags altered S3 encryption settings but the team uses CloudTrail alone as its continuous current-state configuration inventory. Which observed outcome would most convincingly validate the fix?
- A test export contains permitted pseudonyms and omits restricted raw fields.
- A controlled bucket configuration change is detected with historical and current evidence.
- A controlled state machine execution invokes only its intended function.
- The user obtains aggregate headcount without querying raw compensation fields.
Explanation
Correct Answer
B:A controlled bucket configuration change is detected with historical and current evidence. This verifies the result of use AWS Config resource history/rules alongside CloudTrail API events as appropriate.
Incorrect Answers
A:This would verify another situation: A DynamoDB-backed application needs to redact sensitive values before exporting analytics but relies only on table encryption. It does not validate the current case.
C:This would verify another situation: A Step Functions state invokes a Lambda using a role that trusts EC2 instead of the Step Functions service. It does not validate the current case.
D:This would verify another situation: A Lake Formation table exposes a private salary column through a broad Athena SELECT grant when an analyst needs only aggregated headcount. It does not validate the current case.
AWS Certified Data Engineer - Associate DEA-C01 Practice Tests