Google Cloud Platform Services Overview: What to Use First

Google Cloud Platform, usually shortened to GCP or Google Cloud, includes hundreds of services. That is useful once you know what you are building, but it can be hard to navigate when you are just trying to choose the right starting point.

This page is a practical map of the main Google Cloud service families. It is not trying to list every product. The goal is to help you decide which service to look at first for common workloads: virtual machines, containers, storage, databases, analytics, messaging, AI, security, and operations.

Source check – 1 June 2026: Google’s products at a glance page groups current services across AI and ML, application hosting, compute, data analytics, databases, networking, observability, security, storage, and cross-product tools. Check the official product page and launch stage before using a service for production. Google’s service-specific terms explain that pre-GA offerings such as Preview, Alpha, Beta, Early Access, or Experimental features may not have the same support, SLA, or stability expectations as GA services.

On This Page

• Which Google Cloud service should I use?
• How Google Cloud is organised
• Compute and application hosting
• Storage and databases
• Networking and security
• Data, analytics, and messaging
• AI and machine learning
• Operations and developer tools
• Beginner learning path

Which Google Cloud Service Should I Use?

If you are new to Google Cloud, start with the outcome you need rather than the product catalogue. This table gives you the most useful first service to evaluate for common situations.

Need	Start with	Why
Run a traditional server or lift-and-shift workload	Compute Engine	It gives you full virtual-machine control over operating system, disk, network, and runtime.
Run a container without managing servers	Cloud Run	It is the simplest first choice for containerised web apps, APIs, jobs, and event-driven services.
Run Kubernetes	Google Kubernetes Engine (GKE)	It gives you managed Kubernetes while still exposing the Kubernetes model.
Store files, static assets, backups, or raw data	Cloud Storage	It is durable object storage and is often the first storage service a new project needs.
Use a familiar relational database	Cloud SQL	It provides managed MySQL, PostgreSQL, and SQL Server.
Build a global relational application	Spanner	It is designed for globally distributed relational workloads that need strong consistency.
Use a document database for web or mobile apps	Firestore	It is a managed NoSQL document database with mobile and web application use cases.
Analyze large datasets with SQL	BigQuery	It is Google’s managed analytics warehouse and is often the fastest route to reporting at scale.
Connect services with events	Pub/Sub	It decouples publishers and subscribers for event-driven systems.
Build streaming or batch data pipelines	Dataflow	It handles managed stream and batch processing without running your own processing cluster.
Build, tune, or deploy AI/ML workflows	Vertex AI	It is the main managed platform for model development, deployment, and AI workflows.
Monitor production services	Cloud Monitoring and Cloud Logging	They are the core observability tools for metrics, logs, dashboards, and alerts.
Control access and permissions	IAM	Identity and Access Management is the permission model that protects Google Cloud resources.

How Google Cloud Is Organised

Before choosing services, it helps to understand a few Google Cloud building blocks. These concepts appear across almost every product.

Concept	What it means	Why it matters
Project	A container for resources, APIs, IAM policies, billing, and quotas.	Most learning and test work should start in a dedicated project.
Region and zone	Locations where Google Cloud resources run.	Location affects latency, resilience, availability, and cost.
API enablement	Many Google Cloud services must be enabled before use.	If a command fails unexpectedly, check whether the API is enabled for the project.
IAM	The permission system for users, groups, service accounts, and resources.	Good IAM prevents accidental over-permissioning and production risk.
Billing and quotas	Billing controls spend; quotas limit how much of a resource you can use.	Set budgets and alerts before running experiments that create resources.

Start here: If you are new to Google Cloud, practise in Cloud Shell, install the Google Cloud SDK, and learn the basics of gcloud commands before building a complex environment.

Compute and Application Hosting

Compute is where your code runs. The main decision is how much infrastructure control you want to keep.

Service	Use it for	Best first use case	Watch out for
Compute Engine	Virtual machines, custom server builds, lift-and-shift workloads, legacy apps.	Running Linux or Windows servers with full OS control.	You manage patching, hardening, scaling approach, and VM lifecycle decisions.
Cloud Run	Serverless containers, APIs, web apps, background jobs, event-driven services.	Deploying a containerised app without managing Kubernetes.	Design for stateless workloads and understand request, scaling, and timeout behaviour.
Google Kubernetes Engine (GKE)	Managed Kubernetes clusters for container platforms.	Running workloads that already need Kubernetes primitives such as pods, services, ingress, and operators.	Do not choose Kubernetes just because the app uses containers. Cloud Run may be simpler.
App Engine	Managed platform-as-a-service web apps.	Simple web applications where you want Google to manage most runtime concerns.	Runtime and platform constraints matter. Check whether your app fits the supported model.
Cloud Run functions	Single-purpose functions triggered by HTTP requests or events.	Small automation handlers, webhooks, and glue code.	Keep functions focused. If the app grows into a service, Cloud Run may be cleaner.

For most beginners building a new web service, compare Cloud Run and Compute Engine first. Move to GKE when you actually need Kubernetes. If you are ready for Kubernetes, see this TurboGeek guide to creating a GKE cluster.

Storage and Databases

Storage choices depend on the shape of your data. Object storage, block storage, file storage, relational databases, and NoSQL databases solve different problems.

Service	Use it for	Choose it when
Cloud Storage	Object storage for files, media, backups, logs, data lakes, and static assets.	You need durable storage for objects rather than a mounted file system or relational database.
Persistent Disk / Hyperdisk	Block storage attached to VMs.	You are running Compute Engine workloads that need disk volumes.
Filestore	Managed NFS file shares.	Applications need a shared file system interface.
Cloud SQL	Managed MySQL, PostgreSQL, and SQL Server.	You want a familiar relational database with managed backups, patching, and replication options.
AlloyDB for PostgreSQL	High-performance PostgreSQL-compatible workloads.	You need PostgreSQL compatibility with more advanced performance and scale requirements.
Spanner	Global relational workloads with strong consistency.	You need horizontal scale and multi-region consistency that traditional SQL deployments struggle to provide.
Firestore	NoSQL document storage.	You are building web or mobile apps around document data and application-driven reads/writes.
Bigtable	Wide-column NoSQL data at very high scale.	You have time-series, IoT, personalization, or analytical/operational workloads that need low-latency large-scale access.
Memorystore	Managed in-memory caching.	You need Redis, Valkey, or Memcached-style cache behaviour around an application.

A useful first rule: use Cloud Storage for objects, Cloud SQL for normal relational applications, Firestore for document-style apps, and BigQuery for analytics rather than application transactions.

Networking and Security

Networking connects your services. Security controls who can access them and how they are protected.

Area	Start with	What it does
Private networking	Virtual Private Cloud (VPC)	Defines networks, subnets, routes, and firewall rules for cloud resources.
Internet traffic distribution	Cloud Load Balancing	Routes traffic across backends and supports high availability patterns.
Domain name resolution	Cloud DNS	Managed DNS for public and private zones.
Content delivery	Cloud CDN or Media CDN	Caches content closer to users to reduce latency and origin load.
Private outbound access	Cloud NAT	Lets private resources reach the internet without public IP addresses.
Hybrid connectivity	Cloud VPN or Cloud Interconnect	Connects on-premises networks to Google Cloud.
Access control	IAM	Controls who can do what on which resources.
Encryption keys	Cloud Key Management Service	Manages cryptographic keys for supported Google Cloud services.
Security visibility	Security Command Center	Central view of security findings and posture across Google Cloud.
Web and network protection	Google Cloud Armor	Helps protect web applications and services from DDoS and application-layer attacks.

For early projects, spend time on IAM, billing alerts, firewall rules, and least-privilege service accounts before chasing advanced security tooling. Most production problems start with basic controls being too loose.

Data, Analytics, and Messaging

Google Cloud is strong in analytics and event-driven systems. The important split is between storing application data, moving events, processing streams, and analyzing data.

Service	Use it for	Good first scenario
BigQuery	Managed data warehouse and analytics with SQL.	Analyze logs, product data, billing data, marketing data, or application events.
Pub/Sub	Asynchronous messaging between systems.	Decouple services so one system can publish an event and many consumers can react.
Dataflow	Managed streaming and batch data processing.	Transform incoming events or run ETL pipelines without managing processing servers.
Dataproc	Managed Spark and Hadoop ecosystem workloads.	Run existing Spark/Hadoop jobs in Google Cloud.
BigQuery Data Transfer Service	Scheduled imports into BigQuery.	Bring data from supported SaaS tools or storage systems into BigQuery.
Looker / Looker Studio	Dashboards and business intelligence.	Turn data into reports and reusable business views.

If you are learning, try this sequence: send events to Pub/Sub, write them into BigQuery, then build a simple dashboard or query workflow. That teaches messaging, analytics, permissions, and cost controls in one small project.

AI and Machine Learning

Google Cloud’s AI services now cover a wide range of use cases, from managed model development to ready-made APIs for text, speech, documents, images, and translation. The key is to start with the smallest service that solves the problem.

Need	Start with	Why
Build or deploy ML models	Vertex AI	Managed platform for model development, training, deployment, monitoring, and AI workflows.
Use generative AI in Google Cloud workflows	Vertex AI and Gemini-related services	Central place to evaluate Google’s managed model and agent tooling.
Analyze documents	Document AI	Purpose-built extraction and processing for document workflows.
Analyze images	Cloud Vision API	Pre-trained image analysis without building your own model first.
Transcribe speech	Speech-to-Text	Managed speech recognition for audio workflows.
Generate speech	Text-to-Speech	Managed voice synthesis from text.
Translate content	Cloud Translation	Managed translation API for application workflows.
Use AI on warehouse data	BigQuery AI / BigQuery ML	Keeps analysis close to data already stored in BigQuery.

Do not start by training a custom model unless you need one. For many applications, a managed API, Vertex AI workflow, or BigQuery-based model is faster and easier to operate.

Operations and Developer Tools

These tools help you build, deploy, observe, and manage cloud resources. They are not optional once an environment moves beyond experiments.

Area	Tools	Use them for
Browser-based administration	Google Cloud Console	Creating, inspecting, and troubleshooting resources visually.
Command line	Google Cloud SDK, gcloud, gsutil, bq	Repeatable administration, scripting, and terminal-driven workflows.
Browser terminal	Cloud Shell	Running authenticated commands without setting up a local workstation first.
Builds and CI	Cloud Build	Building, testing, and packaging applications.
Packages and images	Artifact Registry	Storing container images and language packages.
Release orchestration	Cloud Deploy	Managed delivery pipelines for supported deployment targets.
Infrastructure as code	Terraform on Google Cloud, Infrastructure Manager, Config Connector	Managing resources as repeatable configuration instead of manual console changes.
Monitoring	Cloud Monitoring	Metrics, dashboards, alerts, and uptime checks.
Logs	Cloud Logging	Centralised log collection, searching, routing, and retention.
Errors and tracing	Error Reporting, Cloud Trace	Debugging application failures and latency across services.

Existing environments may still contain Deployment Manager templates, but new learners should also look at Terraform on Google Cloud and Infrastructure Manager so they understand current infrastructure-as-code options.

Beginner Learning Path

If you are using this page to plan what to learn first, follow this order. It gives you practical coverage without making you learn every service family at once.

1. Create a test project: Set a budget alert and keep experiments away from production.
2. Use Cloud Shell: Learn how authenticated command-line access works in the browser.
3. Install the Google Cloud SDK: Run basic gcloud commands locally.
4. Create a Cloud Storage bucket: Learn projects, IAM, regions, object storage, and permissions.
5. Deploy one small app: Use Cloud Run for a container app or Compute Engine for a VM-based app.
6. Add a database: Use Cloud SQL for a normal relational application or Firestore for a document-style app.
7. Add logging and monitoring: Create a basic dashboard and alert.
8. Try one data service: Use BigQuery for analytics or Pub/Sub for event-driven messaging.
9. Move into advanced services: Learn GKE, Vertex AI, Spanner, or Dataflow only when your workload actually needs them.

Related TurboGeek Guides

• What is the Google SDK?
• Effortless Infrastructure Management with Google Cloud Shell
• How to Get Started with gcloud Commands on GCP
• Create a Kubernetes Cluster on GCP: A Step-by-Step Guide
• What Is Google Pub/Sub and How Does It Work?
• GCP BigTable: The NoSQL Powerhouse
• How to Migrate from Datastore to Firestore
• Resource Monitoring with StackDriver: Practical Guide

Conclusion

The easiest way to understand Google Cloud is not to memorise every product. Start with the workload: compute, storage, data, networking, security, AI, or operations. Then choose the smallest managed service that solves the problem.

For most beginners, the first useful services to learn are Cloud Shell, Google Cloud SDK, Cloud Storage, Cloud Run, Compute Engine, Cloud SQL, BigQuery, Pub/Sub, IAM, Cloud Monitoring, and Cloud Logging. Once those make sense, the larger platform becomes much easier to navigate.