vaayaa Inclusive Education
Discourse13 min read

Google Cloud Platform Basics

By vaayaa Editorial Team · Subject panel: Hiroshi Tanaka (PhD, Computer Science), Carlos García (Master's, Data Science)

GCP is Google's suite of cloud services for computing, storage, networking, databases, analytics and AI, used on a pay-per-use basis. A beginner can build useful basics through self-paced study of resource structure, identity, networking, security and core compute options.

Key takeaways

  • GCP rents computing, storage, networking, data and AI services on a pay-per-use basis [1][2].
  • Regions and zones decide where workloads run, and deploying across multiple zones is the reliability baseline [3][4].
  • Identity and access control, including least privilege and service accounts, matters as much as the choice of compute service [3][5].
  • Security is shared: Google secures the infrastructure, and the customer secures the applications and data [6][7].
  • Market share figures differ between sources, but the three largest providers are consistently reported near 68 percent combined [8][9][10].
  • Employers look for skills you have used, so build small projects and document them [2].

GCP is Google's set of cloud services for computing, storage, networking, databases, analytics, AI and application development [2]. Instead of buying servers, racks and networking gear, a business rents what it needs and pays for what it uses [1]. A learner who understands how these pieces fit together can read job descriptions, follow a design discussion and contribute to a team's cloud work.

This discourse covers the basics a new learner needs: how the platform is organized, the services that matter most, identity and access, security responsibilities, market position, career roles and a self-paced route to competence. The goal is a clear map rather than a catalog of every product. Google's architecture guidance groups good design into five pillars, which works as a practical checklist for any new project [3].

How GCP is organized: regions, zones and projects

Google runs its cloud from a global footprint of points of presence, data centers, regions and zones, and this layout is meant to support high availability and reliability [4]. Architecture guidance recommends deploying across multiple zones as a baseline. Regional services, such as regional GKE clusters and Cloud SQL high availability, allow automatic failover between zones [3].

Resources are grouped into projects. Training material for this discipline explains how resources are organized, how projects are managed, and how service accounts, permissions and networking components fit into that structure [11]. A new learner should create a personal practice project and keep it separate from any shared environment, since billing and access are tied to projects [12].

Where GCP is used

  • Data analysis and machine learning teams that use BigQuery, Vertex AI and related APIs [13].
  • Migration projects that move existing servers into virtual machines [14].
  • Web services and APIs that scale with incoming requests [3].
  • Enterprise platforms spanning data, AI, security and cloud-native infrastructure, as described in a provider market report [15].

Adoption also depends on governance. A Google Cloud executive quoted in a PYMNTS article argues that cloud should be viewed as a strategic enabler rather than merely a new technology to deploy [16]. This framing helps a learner see that projects, roles and policies matter as much as the technical setup.

Core services: compute, storage, networking and serverless options

Compute Engine provides virtual machines for applications that need full operating system control, such as long-running servers and lift-and-shift migrations [14]. Machine families cover general-purpose, compute-optimized, memory-optimized and GPU workloads, and billing is per second, which suits short-lived instances [14].

Cloud Run fits request-based workloads, GKE Autopilot handles container orchestration, and Compute Engine remains the choice when full control is required [3]. A beginner course lists four compute options to compare: Compute Engine, Kubernetes Engine, Cloud Run and App Engine [11]. Storage options named in the same material include Cloud Storage, Filestore and managed databases [11].

Networking starts with virtual private clouds, subnets, firewall rules, routes and load balancing [17][11]. Learners should understand public and private IP addresses before they design a network, because most access decisions depend on them [17].

Choosing a compute tier

The architecture guidance gives a simple decision rule: use Cloud Run for request-based work, GKE Autopilot for container orchestration, and Compute Engine for full control [3]. Treat this as a starting point, then test it against traffic patterns, team skills and how much of the operating system the application needs.

Opinion For most beginners, managed services are a better first step than raw virtual machines, because they reduce the number of setup decisions. Compute Engine still teaches useful concepts about machine types and disks.

Global cloud market share by provider, as reported in one comparison article citing Synergy Research Group figures0%7.5%15%22.5%30%30%AWS25%Azure13%Google Cloud
Global cloud market share by provider, as reported in one comparison article citing Synergy Research Group figures Fact

Source: vaayaa analysis of cited data
View chart data
AWS30%
Azure25%
Google Cloud13%

Identity and access: IAM, service accounts and least privilege

Identity and Access Management decides who can do what in each project [11]. Architecture guidance says to apply least privilege with IAM and to use organization policies as guardrails, such as disabling external sharing or restricting which services a team can use [3]. Grant roles to people and groups, and give applications their own service accounts rather than sharing personal logins.

A learning sequence for this topic usually covers five areas: authentication, roles, service accounts, data protection and encryption, and legislation and compliance [5]. Authentication is the first control. The learner guide recommends enabling multi-factor authentication in the Google Account security settings before anything else [12].

  • Authentication: confirm who the user is before any access is granted [5].
  • Roles: give each person the narrowest role that covers the task [3].
  • Service accounts: let applications act with their own identity [5].
  • Organization policies: enforce limits across all projects in one place [3].
Three GCP compute options compared for a beginner
AspectCompute EngineCloud RunGKEBasis
Typical exampleLong-running servers with custom OS needs [14]Services that respond to requests [3]Container-based services at scale [3]Fact
Level of controlFull control over the operating system [3][14]Managed, request-based compute tier [3]Container orchestration with an Autopilot option [3]Fact
Primary useVirtual machines for full OS control [14]Request-based workloads [3]Container orchestration [3]Fact
Suggested starting point for a beginnerLearn after networking and IAMLearn early for request-based appsLearn after containers are clearOpinion
Fact rows are drawn from [3] and [14]. The beginner row is editorial opinion, not a sourced finding.

Security responsibilities, shared responsibility and compliance

GCP operates under a shared responsibility model. Google secures the cloud infrastructure, while the customer is responsible for securing applications and data deployed on it [6][7]. The split varies with the service type, whether IaaS, PaaS or SaaS, so a learner should check which party owns each control [6].

Fact Google Cloud secures the infrastructure itself, and users are responsible for the security of applications and data running on it [6][7].

Built-in controls include encryption of data at rest and in transit, Cloud Key Management Service for managing keys, VPC Service Controls for security perimeters and Data Loss Prevention [6]. The architecture guidance adds Security Command Center Premium and zero trust access under BeyondCorp [3].

Common risks include misconfigured permissions, inadequate security settings and outdated software [6]. Limited visibility into customer-level access is another challenge in the IaaS model [7], and inadequate logging and monitoring can leave teams unable to detect incidents [18].

Regulated sectors add legal duties on top of technical design. A banking discussion in one trade article stresses that technology and strategy must align to ensure resilience and compliance [16]. This discourse describes these duties for learning purposes only and does not give legal advice.

Cloud market share by provider in a second comparison articleAWS32%Azure24%Google Cloud12%
Cloud market share by provider in a second comparison article Fact

Source: vaayaa analysis of cited data
View chart data
AWS32%
Azure24%
Google Cloud12%

Market position and what the share figures show

One comparison article attributes a global cloud market share of 30 percent to AWS, 25 percent to Azure and 13 percent to Google Cloud, citing Synergy Research Group figures [8]. A second comparison article gives 32, 24 and 12 percent for the same three providers [9]. A third source puts the three together at roughly 67 percent [10].

Estimate Google Cloud's share of the combined three-provider total is about 18 to 19 percent. This figure is calculated in the vaayaa analysis from the two comparison articles [8][9], not measured independently.

The differences are small, but they show why figures from training blogs need checking. Use these numbers for orientation only, and verify them in the original research before you cite them in a report. The more stable point is that GCP is one of three large providers, and job decisions should rest on local job postings [19].

Enterprise demand is also visible in surveys. A study of 532 web-based interviews with senior executives responsible for public cloud investments, conducted with TechTarget's Enterprise Strategy Group and reported by HCLTech, found that 61 percent cite GenAI as the top use case for public cloud infrastructure [20]. The same study reports that 86 percent of organizations say collaborating with third-party experts accelerates GenAI adoption [20].

Roles, employment outcomes and realistic expectations

A training provider's learning pathways name roles such as Cloud Architect, Cloud Developer, Cloud Infrastructure Engineer, Cloud Network Engineer, Cloud Security Engineer, DevOps Engineer and Google Cloud Engineer [21]. Each role draws on the same base of IAM, networking, compute and monitoring, so the basics carry across titles.

One comparison article reports that GCP roles often come later in a career, after 1 to 2 years of experience with the platform [13]. Another comparison article says GCP has the lowest job volume among the three large providers but pays high salaries in data engineering and ML operations roles [19]. Treat both as claims from single sources, not rules.

Resume honesty matters. Interviewers can usually tell when a candidate only knows the buzzwords, so list GCP only for work you have deployed, configured, monitored or secured [2].

  • Operations roles that need monitoring, alerting and incident handling [18].
  • Data and machine learning roles that use BigQuery and Vertex AI [13].
  • Security roles that apply IAM, organization policies and perimeters [3].

Building competence step by step through self-paced study

Start with a safe setup. The learner guide tells learners to sign in with a Google Account they control, enable multi-factor authentication and review trial terms before enabling billing. It also warns against using a corporate billing account unless the organization has approved the training spend [12].

Next, study the console and Cloud Shell to see how resources are managed, then move through resource structure, IAM, networking, compute, storage and managed or serverless services in that order [11][21]. General IT knowledge about servers, networks and applications is helpful but not required [11], and no college degree is needed [17].

Assumption A beginner studies part-time, so the pathway below is measured in steps rather than calendar time. Adjust the pace to the hours you can give each week.

Competence grows when you write down what you built. A learner can create a small project with a storage bucket, a virtual machine and a firewall rule, then record each decision and why it was made [22]. Being able to explain that record to a colleague is the clearest sign that a skill is learned.

Original analysis

vaayaa analysis: comparing two market share figures

  1. Take the AWS, Azure and Google Cloud shares from the first comparison article: 30, 25 and 13 percent [8].
  2. Take the same three providers from the second article: 32, 24 and 12 percent [9].
  3. Subtract the second set from the first for each provider: AWS is 2 points lower, Azure is 1 point higher and Google Cloud is 1 point higher.
  4. Add each set: 30 plus 25 plus 13 equals 68, and 32 plus 24 plus 12 equals 68.
  5. Divide Google Cloud's share by the combined total in each set: 13 divided by 68 is about 19 percent, and 12 divided by 68 is about 18 percent.

The two sources disagree on how the market is split, by up to 2 points per provider, but they agree that the three providers hold about 68 percent combined [8][9]. Within that group, Google Cloud holds about 18 to 19 percent. This is an estimate built from two blog figures, not an independent measurement.

A self-paced pathway for a beginner

  1. 1

    Set up safely

    Sign in with a Google Account you control, enable multi-factor authentication and confirm that billing use is approved [12].

  2. 2

    Learn the structure

    Study organizations, projects, regions and zones while using the console and Cloud Shell [11][4].

  3. 3

    Learn identity

    Practice IAM roles, least privilege and service accounts before you create shared resources [3][5].

  4. 4

    Learn networking

    Study virtual private clouds, subnets, firewall rules and routes, and the difference between public and private IP addresses [17].

  5. 5

    Compare compute

    Match Compute Engine, Cloud Run and GKE to one workload you can describe in plain terms [3][14].

  6. 6

    Build and document

    Create a small storage-based project and write a short record of each decision you made [22].

Worked example

Preparing an image dataset in Cloud Storage

  1. Assume a project holds 1,200 labeled images (assumption for this calculation only).
  2. Assign 80 percent to training: 1,200 times 0.8 equals 960 images.
  3. Assign the remaining 20 percent to testing: 1,200 minus 960 equals 240 images.
  4. Enable the Cloud Storage API for the project, as the Keras example does [22].
  5. Copy the local folders to one bucket with the gsutil copy command, using a bucket name you choose [22].
  6. Check that the bucket holds 960 training files and 240 test files, a total of 1,200.

The dataset is stored in one bucket with a clear 960 and 240 split, ready for a training job. The numbers are assumed to show the arithmetic, not to report a real benchmark.

Practical checklist for getting started

  • Secure your Google Account with multi-factor authentication before anything else [12].
  • Confirm that your organization has approved any billing use [12].
  • Map the resource hierarchy of one practice project on paper [11].
  • Write a least-privilege role plan for each person and service account [3].
  • Compare Compute Engine, Cloud Run and GKE for one workload you can describe [3][14].
  • Record which party owns each security control in your practice project [6].
  • Keep a written record of work you configured so you never list unused tools [2].

Frequently asked questions

Do I need a computer science degree to learn GCP?

No. A college degree is not a prerequisite for a foundational beginner program. Learners should have basic IT knowledge and an interest in cloud and machine learning [17]. Familiarity with servers, networks and applications is helpful but not necessary [11].

Do I need to know programming before I start?

A basic understanding of shell scripting and SQL is listed as helpful, and competency in at least one computer language is helpful though not required [17]. Start with the concepts of projects, IAM and networking, then add scripting as your work requires.

Which GCP service should a beginner learn first?

Opinion: start with IAM and networking, because nearly every service depends on them. Then compare Cloud Run and Compute Engine for a simple workload. Managed services reduce setup decisions, which helps early learners focus on concepts [3].

Is GCP better than AWS for jobs?

The sources do not agree on one answer. One article says AWS has more jobs and GCP leads in AI and ML tools [13]. Another says GCP has the lowest job volume but high salaries in some data roles [19]. Check local job postings before deciding.

Who is responsible for security in GCP?

Under the shared responsibility model, Google secures the cloud infrastructure, and the customer secures applications and data deployed on it [6][7]. Built-in tools such as Cloud KMS, VPC Service Controls and Data Loss Prevention help customers meet that duty [6].

Can GCP skills be used in regulated industries?

Yes, but regulated work requires planning for resilience and compliance from the start, not as an add-on [16]. This answer covers learning and careers only. For legal or compliance requirements, a learner should consult the organization's qualified advisers.

Conclusion

GCP is a set of connected services, and its basics come down to four things: how resources are organized, how identity and access are controlled, which compute option fits a workload, and who is responsible for security. A learner who understands these areas can read job descriptions and join a cloud team's discussions with confidence.

Build your learning in order, keep a written record of what you configured, and be careful with market claims from single sources. The most useful next step is a small, documented project that shows you can make and explain real decisions [2].

Learning sources

References

  1. learn.g2.com. GCP Security: What It Is, Challenges, and Best Practices.
  2. ituonline.com. What Is GCP and Why Google Cloud Skills Are Rising Fast in Job Postings - PublishedMarch 31.
  3. github.com. alpha-engineer/plugins/roles/role-architect/skills/gcp-architect/references/gcp-platform-and-compute.md at main · rnavarych/alpha-engineer - Skip to content.
  4. cntxt.academy. Cloud DevOps Engineer - Google Cloud Fundamentals: Core Infrastructure - 1 | CNTXT Academy.
  5. platform.qa.com. Designing a GCP Security Infrastructure Lesson | QA Learning Platform.
  6. prosperops.com. Google Cloud Security: Best Practices and Tips.
  7. incentro.com. Google Cloud Security: Safeguard Your Business Data and Applications.
  8. tisatech.in. AWS vs Azure vs Google Cloud: Which Cloud Platform Should You Learn in ?.
  9. nexsonitacademy.com. AWS vs Azure vs Google Cloud - Career Comparison.
  10. citadelcloudmanagement.com. AWS vs Azure vs GCP: Which Cloud Platform Should You Learn First in ?.
  11. bittnet.ro. Google Cloud Fundamentals: Core Infrastructure - Bittnet Training.
  12. github.com. Learner Guide.
  13. abctraining.in. Google Cloud vs AWS for Sangli IT Students | ABC Trainings.
  14. notes.kodekloud.com. Summary Understanding compute Persistant disk and firewall - KodeKloud.
  15. everestgrp.com. Everest Group Research Portal.
  16. pymnts.com. Google Cloud: There Are Two Reasons Banks Need to Move to the Cloud | PYMNTS.com - It’s a space where technology and strategy must align to ensure resilience and compliance, and it sits against an evolving and d....
  17. services.google.com. Google Cloud.
  18. sentinelone.com. Google Cloud Security Issues: Key Concerns - Google Cloud Security Issues: Key Concerns.
  19. abctraining.in. AWS vs Google Cloud vs Azure: Which to Learn.
  20. hcltech.com. Google Cloud & GenAI: Trends, Adoption & Business Impact | HCLTech - Overview.
  21. trainocate.com. Google Cloud Platform(GCP) Fundamentals | Certification & Training | Trainocate Lanka.
  22. omi.me. How to Integrate Keras with Google Cloud Platform.
Hourly Active: 40,067 Learners