Metrics

5 minutes 5 Questions

Metrics are the fundamental concept in Amazon CloudWatch that represents a time-ordered set of data points. They are used to track and monitor the performance and utilization of AWS resources, such as EC2 instances, EBS volumes, and RDS databases. You can use CloudWatch to collect and analyze these…

Guide: AWS CloudWatch Metrics

What is AWS CloudWatch Metrics?
AWS CloudWatch Metrics is a scalable AWS service that allows you to collect, view, and analyze metrics for your AWS resources and applications in real-time. It helps you gain system-wide visibility into resource utilization, application performance, and operational health.
Why is AWS CloudWatch Metrics important?
AWS CloudWatch Metrics is important because it allows you to monitor your applications, collect and track metrics, collect and monitor log files, and respond to system-wide performance changes. You can set alarms to notify you when specific thresholds have been breached, and every metric data point can be broken down into a minute or even seconds of granularity.
How does AWS CloudWatch Metrics work?
AWS CloudWatch Metrics works by collecting raw data points from your AWS resources and applications, and turning them into actionable, easy-to-interpret graphs and statistics. It provides you with system-wide visibility into resource utilization, operational performance, and demand patterns-including detailed usage metrics for each service you are provisioned with whether it be EC2 instances, DynamoDB tables, or RDS DB instances.
Exam Tips: Answering Questions on Metrics
1. Make sure to understand the difference between standard and high-resolution metrics: Standard metrics have a minimum granularity of one minute while high-resolution metrics can go down to one second.
2. Keep in mind the different statistic types that CloudWatch provides including Average, Sum, Minimum, Maximum, and SampleCount.
3. Know the limits of CloudWatch such as the fact that data points with a period of less than 60 seconds are available for 3 hours.
4. Remember that metrics are kept for 15 months, so you can retrieve historical data.
5. Finally, understand how to set alarms based on metric thresholds to alert you of potential issues.

Test mode:

Exam (Timed)

Practice (With explanations)

Start practice test

Question 1

You have been asked to create an alarm for when an RDS instance's CPU utilization is above 75% for two consecutive periods of 15 minutes. Which combination should you use?

Evaluation Period: 1, Period Length: 30 minutes Evaluation Period: 4, Period Length: 60 minutes Evaluation Period: 2, Period Length: 60 minutes Evaluation Period: 2, Period Length: 15 minutes

Correct Answer: Evaluation Period: 2, Period Length: 15 minutes

The correct answer requires 2 consecutive evaluation periods of 15 minutes each. Only the correct_answer covers that requirement.

Question 2

An application is experiencing sporadic spikes in end-to-end latency. AWS X-Ray traces show the delay occurs within a specific AWS Lambda function. Which CloudWatch metric should you check first to confirm whether the function’s execution time is causing the latency?

Request per minute Billed time of the Lambda function Duration of Lambda function Concurrent requests

Correct Answer: Duration of Lambda function

When X-Ray pinpoints latency to an AWS Lambda function, the most direct CloudWatch metric to verify whether the function itself is causing the delay is the Lambda Duration metric. Duration measures how long the function takes to execute your code per invocation. Spikes in Duration (especially at p95/p99) that correlate with end-to-end latency spikes confirm the function’s execution time as the root cause. While request rate and concurrency can explain load-related factors, they do not directly indicate execution latency. Billed time is primarily for cost and can include rounding, making it less diagnostic than raw Duration for performance analysis.

Question 3

In Amazon API Gateway (REST APIs), you need to monitor the total time from when API Gateway receives a request from an external client until it sends the response back to that client (as observed by API Gateway). Which CloudWatch metric should you use?

Lambda Function Execution Time Load Balancer Request Processing Time API Gateway Request Latency API Gateway End-to-end Latency

Correct Answer: API Gateway End-to-end Latency

API Gateway Latency is the CloudWatch metric that measures the total time between when Amazon API Gateway receives a request and when it returns a response to the client. This represents the end-to-end duration as observed by API Gateway, including integration time and API Gateway overhead (transforms, auth, etc.), but excluding client-side network time before the request reaches API Gateway. In contrast, IntegrationLatency measures only the backend integration time, Lambda Duration measures just the Lambda function execution, and load balancer timing metrics pertain to ALB/NLB and not API Gateway.

🎓 Unlock Premium Access