AWS Certified Solutions Architect Professional SAP-C02 Practice Question
A company runs dozens of microservices that power its e-commerce site. Each microservice is deployed as an Amazon ECS service on AWS Fargate behind an Application Load Balancer, and the services are spread across multiple AWS accounts in the same AWS Organizations organization. During flash-sale events, customers occasionally experience elevated end-to-end latency and 5xx errors, but the operations team cannot quickly determine which microservice is responsible.
The architecture team wants an AWS-native monitoring solution that will:
- Provide a single, real-time view of end-to-end request traces, error rates, and latency across all microservices
- Display per-task CPU and memory utilization for the Fargate services alongside the traces
- Work across the existing multi-account environment without requiring management of third-party APM agents or custom self-hosted dashboards
Which approach meets these requirements with the LEAST operational effort?
Install the CloudWatch unified agent on every Fargate container, turn on CloudWatch RUM for each web page, and build separate CloudWatch dashboards in each account to correlate metrics, logs, and traces.
Enable AWS CloudTrail Insights in every account, export the events to Amazon S3, run nightly queries with Amazon Athena, and configure anomaly-detection alarms on the query results to detect performance problems.
Designate a CloudWatch monitoring account, enable CloudWatch cross-account observability, instrument the microservices with AWS X-Ray (or the AWS Distro for OpenTelemetry), and use CloudWatch ServiceLens to analyze correlated metrics, logs, and traces from all accounts.
Stream CloudWatch metrics to Amazon Timestream by using CloudWatch Metric Streams, ship Application Load Balancer access logs to Amazon OpenSearch Service, and create performance dashboards in OpenSearch Dashboards for correlation.