Google Associate Cloud Engineer - Practice Test 2
You manage a Google Kubernetes Engine (GKE) cluster for your organization, hosting various non-production workloads for different teams. Your Machine Learning (ML) team requires access to Nvidia Tesla P100 GPUs for model training. You need to implement this solution while minimizing both effort and cost. What is the most appropriate action to take?
Adding a new, GPU-enabled node pool allows specific workloads requiring GPUs to utilize them without impacting other workloads or incurring unnecessary costs for non-GPU-dependent applications. The nodeSelector ensures that ML team pods are scheduled only on the nodes within this new GPU-enabled node pool. This approach optimizes resource allocation and cost efficiency.
You manage a Google Kubernetes Engine (GKE) cluster for your organization, hosting various non-production workloads for different teams. The Machine Learning (ML) team requires access to Nvidia Tesla P100 GPUs for model training. You need to implement this solution while minimizing both effort and cost. What is the most appropriate action to take?
Adding a dedicated GPU-enabled node pool to the existing GKE cluster is the most cost-effective and efficient solution. This allows the ML team to utilize GPUs without impacting other workloads or incurring the higher costs of recreating the entire cluster or managing a separate cluster. The nodeSelector ensures that ML workloads are scheduled only on the GPU nodes.
You are tasked with deploying a new Enterprise Resource Planning (ERP) system on Google Cloud. This application is designed to keep its entire database in-memory for optimal performance and rapid data access. Which Google Cloud Compute Engine configuration is most appropriate for this requirement?
For in-memory databases like those often used in ERP systems (e.g., SAP HANA), memory-optimized machine types are crucial. The M1 machine type is specifically designed for workloads requiring very large amounts of memory, making it the most appropriate choice for an application that holds its full database in-memory. Other options like preemptible instances, GPUs, or local SSDs do not directly address the primary need for high memory capacity.
You have created two Google Kubernetes Engine (GKE) clusters using the `gcloud container clusters create` command: `prod-cluster` and `dev-cluster`. `prod-cluster` is a standard cluster, while `dev-cluster` is an Autopilot cluster. When you execute `kubectl get nodes`, you only see the nodes belonging to `prod-cluster`. Which sequence of commands should you run to view the node status for `dev-cluster`?
The `gcloud container clusters get-credentials` command is used to update your local `kubeconfig` file with the necessary credentials and endpoint information for a specific GKE cluster. This action sets the current context for `kubectl` to the specified cluster. Once the context is set to `dev-cluster`, subsequent `kubectl` commands, such as `kubectl get nodes`, will operate against that cluster.
A healthcare provider stores patient medical images in an on-premises data center. They want to leverage Google Cloud Storage for archival purposes and require an automated process to upload new images from their on-premises system to Cloud Storage. Which solution should be implemented to achieve this?
Option 2 provides a direct and efficient method for synchronizing on-premises data with Cloud Storage. The `gcloud storage rsync` command is designed for this purpose, and scheduling it with a cron job ensures automated, recurring uploads. This approach is practical for integrating existing on-premises systems with cloud storage.