Google Professional Cloud Developer Question 307
Select 3Google Cloud PlatformYou are deploying a new feature for your web application hosted on Google Cloud and want to run an A/B test to compare the performance of two versions of the feature. You plan to use Google Cloud Load Balancer to direct traffic to different versions. Which of the following steps should you take to set up the A/B test correctly?
- A
Configure URL maps in Google Cloud Load Balancer to direct traffic based on specific query parameters or paths.
- B
Use Cloud Monitoring with custom metrics to analyze the performance of each version.
- C
Deploy both versions of the feature on separate backend services and configure traffic splitting rules in Load Balancer.
- D
Store A/B test results in a NoSQL database like Firestore for real-time analysis.
- E
Implement percentage-based traffic splitting directly in the frontend code of the application.
Show answer and explanation
Correct answers: A, B, C
Explanation
A/B testing using Google Cloud Load Balancer involves setting up routing rules through URL maps, deploying the feature versions on separate backend services, and configuring traffic splitting rules for controlled distribution. Cloud Monitoring ensures you can analyze the results effectively, making these steps critical for a successful A/B test. Storing results or implementing frontend-based traffic splitting is not necessary for the core setup.
- A. Correct.
Configuring URL maps in Google Cloud Load Balancer allows you to set up routing rules, enabling traffic to be directed to different versions based on specific parameters, which is essential for A/B testing.
- B. Correct.
Cloud Monitoring with custom metrics is necessary to gather and analyze the performance of each version, helping to determine which version performs better.
- C. Correct.
Deploying both versions on separate backend services and configuring traffic splitting rules is a standard approach to A/B testing, as it ensures controlled and reliable traffic distribution.
- D. Incorrect.
Storing A/B test results in a NoSQL database like Firestore is not required for setting up the test itself but could be useful for long-term storage and analysis after the test.
- E. Incorrect.
Implementing percentage-based traffic splitting in the frontend is not a recommended practice as it adds unnecessary complexity and lacks the reliability of server-side traffic splitting.