Google Professional Machine Learning Engineer Question 164
Select 2Google Cloud PlatformYou are building a machine learning pipeline on Google Cloud and want to integrate your model training code stored in a source code repository with your CI/CD pipeline. You need to ensure seamless version control, automated testing, and deployment while maintaining compatibility with Google Cloud services. Which of the following approaches should you take?
- A
Use Cloud Source Repositories to host your code and configure Cloud Build triggers for automated CI/CD.
- B
Integrate GitHub or Bitbucket with Cloud Build by setting up webhook-based triggers.
- C
Manually download the code from your source repository and upload it to Google Cloud Storage before triggering the pipeline.
- D
Use Artifact Registry to directly store the source code and configure CI/CD pipelines to fetch code from there.
- E
Leverage Google Cloud Deploy to fetch the source code directly from Cloud Source Repositories for model deployment.
Show answer and explanation
Correct answers: A, B
Explanation
To integrate code source repositories with CI/CD pipelines on Google Cloud, Cloud Source Repositories or external repositories like GitHub or Bitbucket can be used. Both options provide seamless integration with Cloud Build for automated testing and deployment. Using manual processes or incorrect services, like Artifact Registry or Cloud Deploy, does not align with best practices or intended use cases for CI/CD pipelines.
- A. Correct.
Cloud Source Repositories is a fully-managed service that integrates seamlessly with Google Cloud services. By configuring Cloud Build triggers, you can automate the CI/CD process for your machine learning pipeline.
- B. Correct.
GitHub and Bitbucket are popular external source code repositories, and Google Cloud allows seamless integration with these platforms using webhook-based triggers in Cloud Build to automate the CI/CD pipeline.
- C. Incorrect.
Manually downloading code and uploading it to Google Cloud Storage is a highly inefficient and error-prone approach. It does not align with best practices for CI/CD automation.
- D. Incorrect.
Artifact Registry is intended for storing build artifacts like container images, packages, or binaries, not for storing source code. It is not suitable for direct integration of source code with CI/CD pipelines.
- E. Incorrect.
Google Cloud Deploy is a continuous delivery service for deploying applications, but it does not directly fetch source code from repositories. It works with artifacts like container images or manifests created during the CI process.