A daily ETL sometimes finishes late, leaving downstream analytical tables stale.
Detect the delay and identify which job stage failed or slowed down.
Which two monitoring choices are appropriate?
Set CloudWatch alarms on pipeline metrics and use structured logs with correlation IDs to trace one execution across stages.
Detailed explanation
Duplicates and prolonged failures can result while the root cause remains unknown.
Duplicates and prolonged failures can result while the root cause remains unknown.
Detection is delayed and the failing processing stage is not identified.
Detection is delayed and the failing processing stage is not identified.
Logs from multiple services can be associated with one run to narrow the failing stage.
Logs from multiple services can be associated with one run to narrow the failing stage.
Signals of abnormal behavior and downstream impact cannot be observed.
Signals of abnormal behavior and downstream impact cannot be observed.
An SLO breach or abnormal run can be detected automatically and routed to notification or retry.
An SLO breach or abnormal run can be detected automatically and routed to notification or retry.
Check official documentation
Compare the following points with the official documentation.
Amazon CloudWatch metrics/alarms、CloudWatch Logs Insights、AWS Glue job metricsの公式説明を確認する。
What to confirm
遅延の検知と原因追跡を、目視確認や無制限再実行と区別できる。
Key points
- CloudWatch alarm
- Structured log
- Correlation ID
- SLO
Notes
- Environment: AWS公式DEA-C01試験ガイドと各サービスの公式ドキュメントの確認
- Confirm current AWS service behavior in the official DEA-C01 documentation.
Foundation review
Detection
Measure freshness, volume, failures, and duration so abnormal runs are visible.
Traceability
Carry execution identifiers through each stage before attempting recovery.