Google Associate Cloud Engineer - Practice Test 3
You have developed an application containerized as a Docker image. You need to deploy this Docker image as a workload on Google Kubernetes Engine (GKE). Which sequence of actions should you take?
To deploy a Docker image on GKE, the image must first be stored in a container registry accessible by GKE, such as Google Container Registry (or Artifact Registry). A Kubernetes Deployment is then used to define and manage the desired state of your application's pods, ensuring they are running and available. A Kubernetes Service, on the other hand, is used to expose your application to network traffic.
The sales department utilizes a Google Cloud project named 'Sales Analytics' with the ID 'sales-analytics-proj'. The marketing department requires a similar set of cloud resources, but these resources must be entirely separate and independently managed from the sales department's resources. What is the most appropriate action to take?
To ensure complete independence and isolation of resources between teams, each team should have its own dedicated Google Cloud project. Projects serve as the fundamental unit of organization and billing, providing clear separation. Reusing project IDs is not possible, and granting roles on an existing project would merge resources and management, violating the independence requirement.
A machine learning team at your company needs a scalable and adaptable platform on Google Cloud to fine-tune large language models using extensive proprietary datasets. You are tasked with designing the optimal solution. Which Google Cloud service combination should you recommend?
GKE with hardware accelerators (like GPUs or TPUs) provides the necessary scalability, flexibility, and orchestration capabilities for fine-tuning large language models. It allows for efficient resource management, distributed training, and integration with MLOps workflows. The other options either lack the specialized capabilities for large-scale ML training or the comprehensive orchestration features required for this use case.
As a team lead, you've provided each of your 10 developers with an individual Google Cloud Project for their personal sandbox environments. You need to be notified if any single developer's sandbox environment exceeds $500 in monthly spending. What is the most efficient way to achieve this?
To monitor individual project spending and receive alerts when a specific project exceeds a threshold, a budget must be set for each individual project. A single budget for all projects would only alert on the aggregate spend, not individual project overages. While BigQuery exports and Data Studio can visualize spending, they do not inherently provide automated alerts based on predefined thresholds like Cloud Billing budgets do.
You have a development project with custom IAM roles defined. You need to replicate these exact custom roles in a new production project with the least administrative effort. Which method should you use?
The `gcloud iam roles copy` command is designed for replicating custom IAM roles efficiently. By specifying the production project as the destination, you can copy roles directly from the development project with minimal steps. Other options involve manual recreation or copying to a broader scope than intended.