Option 1 is the correct answer. CIDR must be configured one for each subnet. This allows you to specify the number of IP addresses assigned within the subnet and the subnet range. In Google Cloud, your VPC is globally configured and has subnets within the region. Therefore, you don't need to assign an IP address range to your VPC, you only assign an IP address range to your subnet.

Option 2 is incorrect. You need to set CIDR for each subnet, not for each region. Regions are not networks, they are geographical areas and are not subject to CIDR.

Option 3 is incorrect. You don't have to assign an IP address range to your VPC, just a subnet.

Option 4 is incorrect. You need to set CIDR for each subnet, not for each zone. Zones are a collection of data centers, not networks, and are not subject to CIDR.

【reference】

https://cloud.google.com/vpc/docs/alias-ip

Option 3 is the correct answer. If you set the status of the image to "DEPRECATED", you can still continue to use the image to create VMs, but you will get a warning. The purpose is to notify the user that the image is obsolete, not to delete the image itself. Because of this, you need to identify that the image will be obsolete while still making it available.

Option 1 is incorrect. Prohibiting access to an image is not appropriate as it will suddenly prevent users from creating VMs using this image. This does not serve the purpose of notifying.

Option 2 is incorrect. Deleting an image is not appropriate as it will suddenly prevent users from creating VMs with this image. This does not serve the purpose of notifying.

Option 4 is incorrect. Increasing the version of the image does not notify the user that the image is obsolete.

【reference】

https://cloud.google.com/compute/docs/images/create-delete-deprecate-private-images

The option 5 project is the wrong component. An App Engine application has at least one service and can hold multiple versions depending on the billing status of the app. Each service consists of multiple versions, these versions are running on multiple instances.

Therefore, options 1, 2, 3 and 4 are the correct components.

【reference】

https://cloud.google.com/appengine/docs/standard/python3/an-overview-of-app-engine

Option 3 is the correct answer. In order to use the GPU in the VM, you need to install the GPU driver in the VM. For NVIDIA GPUs running Compute Engine, you need to use the following NVIDIA driver versions:

[For A100 GPU]

Linux: 450.80.02 or later

Windows: 452.77 or later

[For all other GPU types]

Linux: NVIDIA 410.79 or later

Windows: 426.00 or later driver

Option 1 is incorrect. If you set the GPU to connect to the VM, you do not need to activate it separately.

Option 2 is incorrect. GPU usage and disk space are irrelevant.

Option 4 is incorrect. Since learning can be performed without using the GPU, it is considered that the instance itself is running.

【reference】

https://cloud.google.com/compute/docs/gpus#restrictions

https://cloud.google.com/compute/docs/gpus/install-drivers-gpu#minimum-driver

Option 3 is the correct answer. You can create a VPN from the "VPN" item in "Hybrid Connections" on the console. Here, there are three hybrid connection methods, Cloud Interconnect, Cloud VPN, and peering with Google, which include Dedicated Interconnect and Partner Interconnect, Cloud VPN when the required volume is low, and direct peering and carrier peering, respectively.

Cloud VPN is used when there are few data connections. Cloud VPN connects your on-premises or other public cloud networks to Google VPC securely and at low cost over the Internet via IPSec VPN. It addresses the needs of data bandwidth up to 3.0 Gbps. Cloud VPN's flexible routing options allow you to connect to different VPN gateways using static or dynamic routing. Highly available VPNs offer industry-leading SLAs and guarantee 99.99% uptime.

【reference】 https://cloud.google.com/hybrid-connectivity

Option 2 is the correct answer. The autonomous system number (ASN) is the number used to identify the cloud router on your network. By setting the ASN, you can place and use the cloud router in the network. By entering Google ASN on the router creation screen when setting up a VPN connection, you can set to identify the cloud router on the network.

Option 1 is incorrect. Assigning an IP address does not allow you to assign a unique identifier to the BGP protocol.

Option 3 is incorrect. The MAC address is physically assigned to uniquely identify each carrier on the communication network, and a unique identifier cannot be assigned to the BGP protocol.

Option 4 is incorrect. Assigning a VPC ID does not allow you to assign a unique identifier to the BGP protocol.

[Reference] https://cloud.google.com/network-connectivity/docs/router/concepts/key-terms

Option 3 is the correct answer. App Engine calculates the number of instances needed to handle the current application traffic based on scaling settings such as target_cpu_utilization and target_throughput_utilization. Target_throughput_utilization is set at the same time as max_concurrent_requests to specify the maximum number of concurrent requests to the application before scaling up, and when the number of concurrent requests reaches a value equal to the product of max_concurrent_requests and target_throughput_utilization, the autoscaling scheduler will Launch a new instance.

■ target_throughput_utilization

Optional.

Specify a value between 0.5 and 0.95. The default is 0.6. Used with max_concurrent_requests to specify when concurrent requests launch a new instance.

■ max_concurrent_requests

Optional.

The number of concurrent requests that an autoscaling instance can accept. If this number is exceeded, the scheduler will create a new instance (default: 10, maximum: 80). Used with target_throughput_utilization to specify when a new instance is launched by a concurrent request.

[Reference] https://cloud.google.com/appengine/docs/standard/go/config/appref

Option 1 is the correct answer. A service account is a mechanism that authorizes an application or virtual machine to perform a task. It is possible to grant permissions to the application for which the service account is set to allow writing to the message queue.

Service accounts are special accounts used by applications and computing workloads, not users. Service accounts are managed by Identity and Access Management (IAM).

You can use two types of service accounts: user managed service accounts and Google managed service accounts. A user-managed service account connected to a Compute Engine instance provides credentials to applications running on the instance. These credentials are used by your application to authenticate to the Google Cloud API and authorize access to Google Cloud resources.

Option 2 is incorrect. Google Management Services accounts are created, managed by Google, and automatically assigned to your projects. These accounts represent multiple Google services, and each account has a certain level of access to your Google Cloud project. This is not used to grant VM resource permissions.

Options 3 and 4 are incorrect. There is no project division such as Google managed projects or user managed projects.

Option 4 is the correct answer. You can specify the subnet when creating a firewall by using the'--network' command. It is mandatory to specify a specific subnet when setting up a firewall.

When you create a VPC firewall rule, you specify the VPC network and a set of components that define the behavior of the rule. Components allow you to target specific types of traffic based on the protocol, destination port, source, and destination of the traffic.

[Reference] https://cloud.google.com/vpc/docs/firewalls

Option 1 is the correct answer. Pods are the basic unit of execution for Kubernetes applications. A pod is the smallest and simplest unit in the Kubernetes object model and represents a process running in a cluster. Pods are basically made up of one container, but multiple containers can coexist.

Option 2 is incorrect. A cluster is a collection of nodes that run a containerized application.

Option 3 is incorrect. A single instance on a Kubernetes cluster is not called an "instance".

Option 4 is incorrect. A single instance on a Kubernetes cluster is not called a "volume".