Google Cloud 基础练习题

Compute Engine is Google Cloud IaaS for virtual machines with full OS control.

App Engine is a PaaS service for building and hosting web applications.

Cloud Functions is Google Cloud serverless event-driven compute. It runs code in response to events such as file uploads or Pub/Sub messages, so you pay only while a function executes and never manage servers. The other options require persistent infrastructure or serve batch workloads rather than event-driven functions.

Google Kubernetes Engine (GKE) provides managed Kubernetes clusters. GKE operates the control plane and simplifies upgrades, so teams can run containerized workloads without managing Kubernetes themselves. Cloud Run and App Engine are serverless alternatives, while Compute Engine is unmanaged virtual infrastructure.

Cloud Run executes containers in a fully managed serverless environment.

Cloud Storage is Google Cloud object storage for unstructured data.

Filestore provides managed NFS file storage on Google Cloud. It gives applications a shared POSIX-compliant file system, which is useful when migrating on-premises workloads that expect file shares. Cloud Storage and Persistent Disk do not provide managed NFS shares, and Cloud Spanner is a relational database.

Persistent Disk provides durable block storage attached to Compute Engine VMs.

Cloud SQL offers managed MySQL, PostgreSQL, and SQL Server databases.

Cloud Spanner is a globally distributed relational database with strong consistency.

Firestore is a flexible, serverless NoSQL document database for app development.

Cloud Bigtable is a scalable NoSQL wide-column database for large analytical workloads.

Virtual Private Cloud (VPC) provides isolated private networking for Google Cloud resources.

Cloud Load Balancing distributes traffic across healthy instances and regions.

Cloud CDN uses Google edge locations to cache and deliver content closer to users.

Cloud DNS provides reliable, low-latency domain name resolution. It translates human-readable domain names into IP addresses using Google managed global infrastructure. Cloud NAT handles outbound traffic, Cloud Armor filters attacks, and Cloud CDN accelerates content delivery rather than resolving domain names.

Cloud Armor provides DDoS protection and web application firewall rules.

Cloud VPN provides encrypted tunnels over the public internet. It connects an on-premises network to Google Cloud with IPsec, which is practical when a dedicated private link is not required. Cloud Interconnect uses physical connections, while Cloud NAT and Cloud CDN do not create site-to-site encrypted tunnels.

Cloud Interconnect provides dedicated private connections to Google Cloud. Because traffic does not travel over the public internet, the link offers higher bandwidth and more consistent latency than Cloud VPN. Cloud NAT only translates outbound traffic, and Cloud Armor filters requests instead of providing connectivity.

BigQuery is Google Cloud serverless data warehouse for petabyte-scale analytics.

Dataflow is a managed service for stream and batch data processing pipelines.

Pub/Sub is a scalable event ingestion and delivery service.

Dataproc is a managed service for Hadoop, Spark, Flink, and related clusters.

Vertex AI unifies Google Cloud ML tools for building and deploying models.

Vision AI provides pre-trained image recognition capabilities. Its APIs can detect objects, faces, text, and unsafe content, so developers add image understanding without training custom models. Pub/Sub, Cloud NAT, and BigQuery do not perform image analysis.

Speech-to-Text transcribes audio into text using machine learning. It supports many languages and can return word timestamps, which makes it useful for captions, transcripts, and call analytics. Translation AI converts text between languages, Vision AI analyzes images, and Natural Language AI processes text rather than audio.

Translation AI translates text between many languages. It offers pre-trained models that preserve meaning across language pairs through a simple API. Speech-to-Text converts audio, Vision AI analyzes images, and Cloud DLP protects sensitive data instead of translating text.

Cloud IAM controls who can access which Google Cloud resources.

Cloud KMS manages cryptographic keys for encryption on Google Cloud.

Security Command Center provides security findings, posture, and threat detection.

Cloud Logging stores, searches, and analyzes logs from Google Cloud resources.

Cloud Monitoring collects metrics, dashboards, and alerts for Google Cloud.

Error Reporting aggregates and analyzes application errors. It groups related stack traces, shows crash counts, and can notify teams, which speeds up debugging of large systems. Cloud Logging stores raw logs, Cloud Monitoring tracks metrics, and Cloud Debugger inspects running code, so none of them group application errors the same way.

Cloud Build executes build steps and produces container images. It can be triggered by source changes, so CI/CD pipelines compile code and push artifacts automatically. Artifact Registry stores the images, Cloud Run serves them, and Cloud Storage is object storage rather than a build service.

Artifact Registry stores container images and software packages. It provides a central, access-controlled location for build artifacts and integrates with Cloud Build and other services. Cloud Build creates artifacts, Cloud Run runs containers, and Cloud Storage is general object storage.

Secret Manager stores and manages API keys, passwords, and certificates.

Cloud Build can create CI/CD pipelines that test and deploy applications.

Cloud Deployment Manager uses declarative templates to deploy Google Cloud resources.

Sustained use discounts apply automatically as a VM runs for most of a month.

Committed use discounts offer savings when you commit to one or three years of usage.

Preemptible and spot VMs are cheaper but can be terminated at any time.

The Pricing Calculator estimates monthly costs for Google Cloud services.

Billing budgets track spending and send alerts when thresholds are reached.

A project organizes Google Cloud resources, billing, and IAM permissions.

A region is a specific geographic location where Google Cloud resources run.

A zone is a deployment area within a region; regions contain multiple zones.

Labels are key-value pairs used to organize and filter Google Cloud resources.

Horizontal scaling adds more instances to handle increased demand. Workloads are distributed across additional virtual machines or replicas, which improves capacity without upgrading a single machine. Increasing the size of one instance is vertical scaling, while moving regions or changing billing does not add capacity.

Vertical scaling increases the capacity of an existing resource, such as using a larger machine type.

Cloud Run runs containers in a serverless environment with automatic scaling.

Data Catalog helps discover and manage metadata for data assets.

Cloud Data Loss Prevention (DLP) discovers, classifies, and protects sensitive data.

Identity-Aware Proxy (IAP) uses identity and context to control application access.

VPC Service Controls creates perimeters that protect Google Cloud service data.

Cloud NAT lets private instances reach the internet without public IP addresses.

AlloyDB for PostgreSQL is a fully managed PostgreSQL-compatible database. It provides high availability, automated backups, and scalability for transactional workloads. Firestore is a NoSQL document store, Bigtable is a wide-column database, and Memorystore is an in-memory cache, so none of them are PostgreSQL-compatible.

Memorystore provides managed Redis and Memcached in-memory data stores. It is used for caching and low-latency access patterns so applications can offload repeated reads from databases. Cloud Storage and Cloud SQL are persistent storage services, and Filestore provides file shares rather than in-memory caches.

Memorystore for Redis is a managed Redis service for caching and low-latency data.

Cloud Load Balancing provides global and regional load balancing for HTTP(S) and other protocols.

BigQuery is a serverless data warehouse that scales query capacity automatically.

Compute Engine, App Engine, and Cloud Functions are compute services; Cloud Storage is storage.

Cloud Storage, Filestore, and Persistent Disk are storage services; Load Balancing is networking.

VPC, Load Balancing, and CDN are networking; KMS manages encryption keys.

IAM, KMS, and Security Command Center support security; CDN is content delivery.

BigQuery, Dataflow, and Pub/Sub process data; Cloud DNS resolves domain names.

Vertex AI, Vision AI, and Speech-to-Text are AI services; Cloud Storage is storage.

Cloud Build, Artifact Registry, and Cloud Monitoring support DevOps; Cloud DNS is networking.

Sustained use, committed use, and preemptible VMs reduce compute costs; Cloud NAT is networking.

Compute Engine is Google Cloud IaaS for virtual machines. You choose machine type, operating system, and storage, then manage the instance while Google operates the underlying infrastructure. The blank asks for the service that provides virtual machines, and Engine completes the name Compute Engine.

Cloud Functions runs event-driven serverless code. It executes functions in response to events and scales automatically, so there are no servers to provision and billing is based on execution. The blank asks for the event-driven compute service, and Functions completes the name Cloud Functions.

Google Kubernetes Engine (GKE) provides managed Kubernetes clusters. Google operates the control plane and simplifies upgrades, making container orchestration easier for teams. The missing word is Engine because the full service name is Google Kubernetes Engine.

BigQuery is Google Cloud serverless data warehouse. It runs SQL analytics over very large datasets without provisioning clusters and separates storage from compute. The blank needs Query because the service name is BigQuery.

Cloud Run executes containers in a serverless environment. It scales automatically and charges only for request processing or active instances, so you do not manage servers. Run is the word that completes the service name Cloud Run.

Cloud IAM controls access to Google Cloud resources. It uses roles and policies to grant permissions to users, groups, and service accounts, which supports least privilege. The blank should be IAM because the service is named Cloud IAM.

Compute Engine is IaaS, App Engine is PaaS, Cloud Functions is serverless functions, and Cloud Run is serverless containers.

Cloud Storage stores objects, Filestore provides files, Persistent Disk is block storage, and Cloud SQL is a relational database.

VPC provides private networks, Load Balancing distributes traffic, CDN caches at edges, and VPN creates encrypted tunnels.

BigQuery is a warehouse, Dataflow processes pipelines, Pub/Sub handles messaging, and Bigtable is a NoSQL database.

IAM manages access, KMS manages keys, Security Command Center detects threats, and Cloud Armor protects applications.

Sustained use is automatic, committed use requires a commitment, preemptible VMs are interruptible, and budgets track spending.