AWS Monitoring, Logging, and Disaster Recovery 4 — Questions and Answers
Question 1: A company wants to detect when an EC2 instance's CPU utilization remains above 90% for more than 15 consecutive minutes. Which CloudWatch alarm configuration should be used?
- Set threshold to 90%, period to 900 seconds, evaluation periods to 1
- Set threshold to 90%, period to 300 seconds, evaluation periods to 3 (Correct answer)
- Set threshold to 90%, period to 60 seconds, evaluation periods to 15
- Set threshold to 90%, period to 900 seconds, evaluation periods to 3
Correct answer: Set threshold to 90%, period to 300 seconds, evaluation periods to 3
Setting a 300-second period with 3 consecutive evaluation periods means the alarm triggers only after 15 minutes of sustained high CPU.
Question 2: Which AWS service allows you to query CloudWatch Logs data using standard SQL-like syntax without moving the data to S3?
- CloudWatch Logs Insights (Correct answer)
- Amazon Athena
- Amazon OpenSearch Service
- AWS Glue
Correct answer: CloudWatch Logs Insights
CloudWatch Logs Insights provides an interactive query engine for log data stored directly in CloudWatch Logs groups.
Question 3: An architect needs to implement a multi-region active-passive DR strategy with an RTO of 1 hour and RPO of 15 minutes for an RDS database. Which approach meets these requirements most cost-effectively?
- RDS Multi-AZ with automated backups to S3
- RDS cross-region read replica with automated snapshots every 15 minutes (Correct answer)
- Aurora Global Database with a secondary region
- DynamoDB global tables with point-in-time recovery
Correct answer: RDS cross-region read replica with automated snapshots every 15 minutes
A cross-region read replica can be promoted within an hour while 15-minute automated snapshots satisfy the RPO, making it more cost-effective than Aurora Global Database for non-Aurora workloads.
Question 4: A company needs CloudTrail logs retained for 7 years for compliance. What is the most operationally efficient way to achieve this?
- Configure CloudTrail to send logs to CloudWatch Logs with a 7-year retention policy
- Enable CloudTrail and store logs in an S3 bucket with an S3 Lifecycle policy transitioning to Glacier Deep Archive after 90 days (Correct answer)
- Create a Lambda function that archives CloudTrail logs to Glacier monthly
- Export CloudTrail logs to an RDS database with automated backups
Correct answer: Enable CloudTrail and store logs in an S3 bucket with an S3 Lifecycle policy transitioning to Glacier Deep Archive after 90 days
S3 Lifecycle policies automating transition to Glacier Deep Archive provide durable, low-cost long-term retention with no operational overhead.
Question 5: Which CloudWatch metric is NOT available by default and requires installing the CloudWatch Agent on an EC2 instance?
- CPUUtilization
- NetworkIn
- MemoryUtilization (Correct answer)
- DiskReadOps
Correct answer: MemoryUtilization
MemoryUtilization is an OS-level metric not exposed by the EC2 hypervisor, so it requires the CloudWatch Agent to collect and publish.
Question 6: A solutions architect is designing a DR plan and needs to minimize costs while keeping a warm standby in a secondary region with an RTO under 30 minutes. Which strategy fits best?
- Multi-site active-active
- Backup and restore
- Pilot light
- Warm standby (Correct answer)
Correct answer: Warm standby
A warm standby maintains a scaled-down but fully functional environment in the secondary region, enabling promotion within the 30-minute RTO without the full cost of active-active.
Question 7: An application writes structured JSON logs to stdout in an ECS Fargate task. Which configuration sends these logs to CloudWatch Logs with the least operational overhead?
- Install the CloudWatch Agent as a sidecar container
- Use the awslogs log driver in the ECS task definition (Correct answer)
- Configure a Kinesis Data Firehose delivery stream from the container
- Write logs to an EFS volume and sync with a Lambda function
Correct answer: Use the awslogs log driver in the ECS task definition
The awslogs log driver is natively supported by ECS Fargate and routes stdout/stderr directly to CloudWatch Logs without additional infrastructure.
AWS Monitoring, Logging, and Disaster Recovery 4
A company wants to detect when an EC2 instance's CPU utilization remains above 90% for more than 15 consecutive minutes.
Which CloudWatch alarm configuration should be used?