Blog Post
See the top GCP services ranked by real-world utility. Learn which ten Google Cloud tools you actually need to master for your career and certification exams.
Checking sign-in status...
Google Cloud Platform (GCP) has hundreds of services. You do not need to learn all of them. Most real-world GCP work and the core certification exams (ACE and PCA) focus on the same ten services. Master these, and you cover the majority of what employers test and deploy.
GCP holds roughly 11% of the cloud market, but it is the fastest-growing major provider by percentage. The supply of GCP-certified engineers remains constrained relative to demand, which creates a premium for GCP skills. GCP Professional Cloud Architect ranks among the highest-paying IT certifications, with median total compensation of $162,000 in the US.
This guide covers the ten services that matter, how they map to the exam domains, and what to build with each one.
Salary figures are estimates based on 2026 industry reports and vary widely by location, industry, experience level, and company size. Use them as directional signals, not guarantees.
| # | Service | Category | ACE/PCA Weight |
|---|---|---|---|
| 1 | Compute Engine | Compute | High |
| 2 | VPC | Networking | High |
| 3 | Cloud Storage | Storage | High |
| 4 | Cloud IAM | Identity | Essential |
| 5 | BigQuery | Data | High |
| 6 | GKE | Containers | High |
| 7 | Cloud Run | Serverless | Medium |
| 8 | Cloud SQL | Database | Medium |
| 9 | Cloud Load Balancing | Networking | Medium |
| 10 | Cloud Logging & Monitoring | Operations | High |
Compute Engine provides customizable virtual machines. Even in a cloud known for Kubernetes and serverless, VMs are still used for lift-and-shift migrations, legacy software, and custom database hosting.
What to master: Custom machine types, preemptible VMs for cost savings, managed instance groups (MIGs), and disk types.
Build this: Launch a VM, SSH into it, install Nginx, open port 80 in a firewall rule, and serve a page. Then break it. Close the port and watch the connection fail. Reopen it. This teaches you how networking and compute interact.
GCP networking differs from AWS and Azure. A GCP VPC is global by default, not regional. Subnets are regional. This design allows resources in different regions to communicate without complex peering.
What to master: Global vs. regional subnets, firewall rules applied at the network level, and Cloud NAT.
Build this: Create a VPC with two subnets in different regions. Launch a VM in each. Verify they can communicate without peering. Then add a firewall rule that blocks traffic and observe the failure.
Cloud Storage is the object storage service used for unstructured data, backups, static web assets, and data lake ingestion.
What to master: Storage classes (Standard, Nearline, Coldline, Archive), lifecycle management, and uniform bucket-level access.
Build this: Create a bucket, upload files, set a lifecycle policy to move objects to Coldline after 30 days, and enable uniform bucket-level access. Then try to make a single object public and observe why uniform access blocks it.
Identity is the security perimeter. Cloud IAM controls who and what can access your resources. GCP's resource hierarchy (Organization → Folder → Project → Resource) means IAM inheritance is critical.
What to master: Basic vs. predefined vs. custom roles, service accounts, and policy inheritance.
Build this: Create a service account with Viewer access to a single bucket. Test what it can and cannot do. Add a role at the project level. Observe how inheritance changes its permissions.
BigQuery is GCP's fully managed, serverless data warehouse. Companies choose Google Cloud for BigQuery's petabyte-scale analytics and ML integration.
What to master: How it separates compute from storage, partitioning vs. clustering, and cost control.
Build this: Load a public dataset from BigQuery's public data catalog. Run a query without partitioning. Check the bytes processed. Add a partition filter. Run the same query. Compare the cost.
Google invented Kubernetes. GKE is the managed Kubernetes service used to deploy, scale, and manage containerized applications.
What to master: Standard vs. Autopilot modes, node pools, and container networking.
Build this: Create a GKE Autopilot cluster. Deploy a simple containerized app. Expose it with a LoadBalancer service. Scale the deployment from 1 to 3 replicas. Watch the pods distribute.
Cloud Run lets you deploy containers without managing a Kubernetes cluster. It scales from zero to global traffic automatically.
What to master: Concurrency, cold starts, and integrating with Cloud SQL.
Build this: Containerize a simple Python or Node.js app. Deploy it to Cloud Run. Set concurrency to 1 and observe cold starts. Raise concurrency to 80 and observe the difference.
Cloud SQL provides managed MySQL, PostgreSQL, and SQL Server. Google handles backups, patching, and scaling.
What to master: High availability (regional vs. zonal), read replicas, and Cloud SQL Auth Proxy.
Build this: Create a Cloud SQL instance with a public IP. Connect from a local machine using Auth Proxy. Then remove the public IP and connect from a VM in the same VPC using private IP.
GCP's global network allows Cloud Load Balancing to distribute traffic across regions using a single global IP.
What to master: Global vs. regional load balancers, HTTP(S) vs. TCP/UDP routing, and backend services.
Build this: Deploy two VM instances in different regions. Create an HTTP(S) load balancer with both as backends. Use a single global IP. Test from multiple locations.
Observability is non-negotiable in production. This is the operations suite (formerly Stackdriver) used to track performance and debug errors.
What to master: Log sinks (routing logs to BigQuery or Cloud Storage), uptime checks, and alerting policies.
Build this: Create a log sink that routes specific logs to BigQuery. Query them with SQL. Set up an uptime check on a public endpoint. Create an alerting policy that fires when the endpoint fails.
Reading about these services is not enough. The ACE and PCA exams test whether you can make architectural decisions under constraints. A question might describe a company running BigQuery queries that cost too much and ask which change reduces cost without affecting performance.
ExamOS delivers scenario-based practice that mirrors this format. The Daily Sprint gives you seven questions in seven minutes, adapting to your weak areas automatically. Your Readiness Score tracks accuracy, coverage, and recency across every domain, so you know when you are prepared. Mock Exam simulates the full exam across all domains, weighted exactly as the official syllabus.
Every question follows a Scenario → Constraints → Decision Required structure. You reason through trade-offs, not recall definitions.
The Associate Cloud Engineer (ACE) and Professional Cloud Architect (PCA) exams heavily test these services. The Professional Data Engineer exam focuses deeply on BigQuery and Cloud SQL. ACE is the entry point; PCA is the advanced architecture credential.
A Google Cloud VPC is global by default rather than regional. Subnets are regional, which allows resources in different regions to communicate without complex peering or transit gateways. Firewall rules are applied at the network level, not the subnet level.
Choose Cloud Run for simple, containerized applications that need to scale to zero without cluster management overhead. Use GKE when you require advanced Kubernetes orchestration, custom node pools, or complex microservices.
ExamOS uses the Daily Sprint for short daily scenario practice and a Readiness Score to track progress across every domain. Mock Exam simulates the full exam weighted exactly as the official syllabus. When your Readiness Score crosses the threshold, you know you are ready to book.
Build the GCP skills employers test. Start with a free Daily Sprint on ExamOS and track your Readiness Score as you prepare for ACE or PCA.