Cloud Logging and Kubernetes Management Quiz

Estimating the cost of a Google Kubernetes Engine (GKE) cluster involves using the GCP pricing calculator, considering factors like the number of nodes, average hours and days of use, high IOPs, persistent disk storage, and snapshots.

High Input/Output Operations Per Second (IOPs) are a demanding workload that requires faster storage in a Kubernetes environment.

Persistent disks store data in a Kubernetes cluster, even if the pods stop, requiring a cost assessment.

Capturing copies of the data in persistent disks within Kubernetes for backups or recovery, affecting the cost.

Estimating the cost associated with managing a Kubernetes cluster, including resources, security, and operations.

Making a cloud application accessible to the public through a secure connection (HTTPS) using a public IP address.

A Kubernetes service type that exposes the application on a specific port on each node within the cluster.

Using a Kubernetes Ingress component to manage traffic routing and a Cloud Load Balancer to distribute traffic securely to the containers.

Filtering Compute Engine logs and exporting them to BigQuery's platform-logs dataset using Cloud Logging's export feature reduces storage costs by only storing necessary logs.

Directly define the DaemonSet manifest within the Deployment Manager configuration for the Kubernetes Engine cluster, without using intermediary resources like Compute Engine instances or Runtime Configurators.

BigQuery (platform-logs dataset) is the destination for exporting computed logs in Cloud Logging.

Deployment Manager can manage the creation and configuration of Google Kubernetes Engine (GKE) clusters, as well as resources within the cluster, such as DaemonSets.

A DaemonSet in the kube-system namespace is designed to run on every node within a Kubernetes cluster; this involves running system-required services on every node.

Deployment Manager's Runtime Configurator is not the best choice for deploying Kubernetes resources, as it involves more steps and services than needed.

Using a compute engine instance and a startup script with kubectl to create DaemonSets is not the most efficient approach when deploying within Kubernetes through Deployment Manager.

Metadata in Deployment Manager is a key-value pair approach, and a direct approach for defining resources, therefore it's the recommended method.

Connecting Compute Engine instances in separate GCP projects, each with its own VPC, requires sharing the VPC from one project with the other project's instances.

To enable inter-project VPC communication, both involved projects must be part of the same GCP Organization.

The built-in IAM Project Viewer role grants read-only access to all project resources, suitable for auditors.

To allow an ML team to use Nvidia Tesla P100 GPUs for training in a GKE cluster, use a node pool with GPU-enabled machine types.

The cost of a GKE cluster depends on factors such as node type, storage, and network usage.

You can create custom IAM roles with specific permissions to control access to GCP resources for different users or groups.

Shared VPCs allow sharing a VPC network between multiple GCP projects, facilitating communication and resource sharing.

A GCP Organization provides centralized management, billing, and security for multiple projects within the same organization.