GCP Professional Cloud Architect 练习

Google Cloud resources sit inside projects, projects can be grouped in folders, and folders and projects belong to an organization. The other orders are incorrect.

A project is the base unit that organizes resources, enables APIs, and is linked to billing. VMs, DNS records, and firewall rules are resources inside projects.

Folders group projects and let you apply IAM policies and organization policies hierarchically. They do not store images, manage DNS, or filter traffic.

The organization node is the root that contains folders and projects and supports organization-wide policies. It is not a compute cluster, bucket, or subnet.

A Virtual Private Cloud is a global network that connects resources with subnets, routes, and firewalls. It is not a VM, schema, or billing account.

A region is a geographic area containing multiple zones, and each zone is an isolated facility. DNS zones, folders, and subnets are different concepts.

Multi-region deployments place resources in several regions to improve availability and durability. A single VM or storage class is not a multi-region deployment.

An SLA defines service level expectations such as uptime percentage and credits. It does not define projects, machine types, or DNS records.

High availability uses redundancy, health checks, and failover so service continues during failures. Expensive machines, single zones, and no backups reduce availability.

The organization is the root, folders group projects, projects contain resources, and regions are geographic areas with zones.

Compute Engine provides virtual machines with full control over the OS. App Engine is PaaS, Cloud Functions is serverless, and BigQuery is analytics.

GKE is a managed Kubernetes service for containers. Compute Engine is VMs, Cloud Storage is objects, and Cloud Load Balancing routes traffic.

Cloud Functions is serverless event-driven compute. Compute Engine requires VM management, Bare Metal is physical, and Cloud SQL is a database.

Cloud Run runs stateless containers and scales automatically. Compute Engine is VMs, Cloud Storage is objects, and Cloud DNS resolves names.

Managed instance groups use instance templates, autoscaling, and health checks to manage identical VMs. Unrelated VMs and network objects are not managed groups.

Machine types define vCPU and memory for VMs. Project names, routes, and storage classes are different.

App Engine is a managed PaaS for web apps. Compute Engine is IaaS, Cloud Functions is serverless, and Cloud Spanner is a database.

Spot VMs offer large discounts but can be reclaimed, so they suit fault-tolerant and batch workloads. They are not guaranteed or unlimited.

Unmanaged instance groups contain VMs that you manage independently and do not support autoscaling or templates. Managed groups handle identical VMs.

Compute Engine provides VMs, GKE manages Kubernetes, Cloud Run scales containers, and App Engine hosts web apps.

Cloud SQL offers managed MySQL, PostgreSQL, and SQL Server. Firestore and Bigtable are NoSQL, and Cloud Storage is object storage.

Cloud Spanner combines relational schemas, SQL, and global scale. Firestore and Bigtable are NoSQL, and Cloud Functions is compute.

Firestore is a flexible NoSQL document database. Cloud Spanner and Cloud SQL are relational, and Cloud Storage is objects.

Bigtable is a scalable NoSQL wide-column database with low latency. Cloud SQL is relational, App Engine is compute, and Cloud CDN caches content.

BigQuery is a serverless, highly scalable data warehouse with SQL analytics. Spanner is transactional, Storage is objects, and Filestore is file storage.

Cloud Storage is object storage for arbitrary files. Cloud SQL is relational, Firestore is documents, and Bigtable is wide-column NoSQL.

Filestore provides managed NFS file storage. Cloud Storage is objects, BigQuery is analytics, and Cloud Functions is compute.

Dataproc provides managed Hadoop and Spark clusters. Cloud Run and App Engine run applications, and Cloud CDN caches content.

Cloud Storage offers scalable object storage with storage classes and lifecycle rules. The other services are databases or analytics.

Cloud SQL is relational, BigQuery is a warehouse, Firestore is document NoSQL, and Cloud Storage is object storage.

Subnets are IP ranges within a VPC where resources are placed. VMs, firewall rules, and DNS records are different objects.

Cloud Load Balancing distributes traffic across instance groups and regions. DNS resolves names, NAT provides outbound access, and Interconnect is private connectivity.

Cloud VPN creates encrypted IPsec tunnels over the internet. Interconnect is a private physical connection, CDN caches content, and Armor protects web apps.

Cloud Interconnect provides dedicated or partner-managed physical connections. VPN uses encrypted internet tunnels, NAT provides outbound access, and DNS resolves names.

Cloud NAT lets private VMs reach the internet for outbound connections without public IPs. Load balancers route traffic, CDN caches content, and DNS resolves names.

Firewall rules allow or deny traffic based on source, destination, ports, and protocols. A VPC is the network, buckets store objects, and machine types define VMs.

VPC peering connects two VPCs so resources can communicate privately. Subnets, DNS zones, and billing are separate concepts.

Shared VPC lets a host project share its VPC subnets with service projects for centralized network administration. It is not public IPs, DNS, or backup.

Cloud Load Balancing includes global HTTP(S) load balancers that route traffic to backends worldwide. VPN, Interconnect, and NAT serve connectivity.

Load Balancing distributes traffic, VPN creates encrypted tunnels, Interconnect provides physical connectivity, and NAT gives private VMs outbound access.

Cloud IAM controls who can do what on resources through roles and policies. DNS, Storage, and Logging have different purposes.

IAM roles group permissions that can be granted to principals. VMs, buckets, and subnets are resources, not roles.

Service accounts are identities for applications to authenticate and access Google Cloud APIs. They are not user logins or network objects.

Secret Manager stores and manages access to secrets. Cloud Storage is objects, BigQuery is analytics, and Cloud Functions is compute.

Cloud KMS creates, rotates, and controls encryption keys. IAM manages access, NAT provides outbound connectivity, and DNS resolves names.

Cloud Armor protects applications with WAF rules and DDoS defense. CDN caches content, Interconnect is private connectivity, and VPN creates tunnels.

Cloud Security Scanner finds vulnerabilities in App Engine and Compute Engine web apps. DNS, Storage, and BigQuery do not scan applications.

Organization policies enforce constraints such as allowed locations or resource limits across the hierarchy. They are not firewalls, images, or DNS records.

Cloud Audit Logs record admin, data access, and policy events. DNS, CDN, and NAT do not provide audit trails.

IAM manages permissions, KMS manages keys, Secret Manager stores secrets, and Cloud Armor protects web apps.

Autoscalers add or remove instances in managed groups based on utilization or metrics. They do not manage buckets, DNS, or keys.

Health checks test instances and determine whether they should receive traffic. Backups, firewalls, and project names are different.

Backups create recoverable copies of data and systems. VMs, DNS zones, and load balancers are infrastructure components.

Disaster recovery defines recovery objectives and procedures for restoring systems after major failures. Backups, health checks, and cost reports are narrower tools.

Cloud Monitoring collects metrics and supports dashboards and alerting. Storage, DNS, and Interconnect do not monitor systems.

Cloud Logging ingests, stores, and queries logs. BigQuery is analytics, Firestore is documents, and CDN caches content.

Uptime checks test endpoints from multiple locations and alert when they fail. Firewalls, snapshots, and quotas are unrelated.

Cloud Monitoring alerting policies trigger notifications and actions based on conditions. DNS, storage, and peering do not alert on metrics.

Rolling deployments update instances in stages while keeping the service available. Deleting everything, creating projects, or disabling monitoring are not rolling updates.

Autoscalers adjust instance counts, health checks verify readiness, Cloud Monitoring watches metrics, and Cloud Logging stores logs.