GCP
Beginner 13 min read GCP

The GCP Service Catalog

Find any Google Cloud service by its family

Google Cloud organizes into compute, storage, databases, networking, data/ML and security families. GKE, Compute Engine, Cloud Storage, BigQuery, VPC and IAM come first.

GCPCompute EngineGKECloud StorageBigQueryVPCservice catalog
Prerequisites: What is Cloud Computing?
Before we start

What you will learn

  • Group GCP services into core functional families
  • Map a typical web app to GCP services
  • Navigate the console, gcloud and docs
In 60 seconds

GCP families: compute (Compute Engine, GKE, Cloud Run, Cloud Functions), storage (Cloud Storage, Persistent Disk, Filestore), databases (Cloud SQL, Spanner, Firebase/ Firestore, Bigtable), networking (VPC, Cloud Load Balancing, Cloud DNS), data/ML (BigQuery, Dataflow, Vertex AI), security (IAM, KMS, security command center).

Introduction

What is this?

Google Cloud Platform runs in Google data centers on the same architecture that powers Google. Core services: Compute Engine (VM), GKE (managed Kubernetes), Cloud Run (serverless containers), Cloud Functions (FaaS), Cloud Storage (object), Persistent Disk (block), Filestore (file), Cloud SQL (managed Postgres/MySQL), Spanner (globally consistent SQL), Firestore (document NoSQL), BigQuery (serverless warehouse), VPC (networking), Cloud Load Balancing, Cloud IAM, Cloud KMS. Projects are the top-level container.

Motivation

Why do you need it?

  • GCP inherited Google's data and networking engineering — BigQuery, Spanner, and GKE lead their categories.
  • Multicloud skills: GCP's IAM and VPC model map cleanly to AWS/Azure once you learn the vocabulary.
  • Pricing is per-second and per-1000-GB, with sustained-use discounts — useful in interviews.
  • The console is clean and gcloud is scriptable, making exploration fast.
Think about it

Mental model

Imagine

A research lab with departments

Think of

GCP service families

Each department (compute, storage, data/ML) has sub-labs. You ask the computing department for a server, the storage department for a bucket — the org chart tells you where the tool lives.

Mechanics

How it works

  • Compute: Compute Engine (VMs), GKE (K8s), Cloud Run (containers-on-demand), Cloud Functions (functions).
  • Storage: Cloud Storage (buckets with classes), Persistent Disk (zonal/regional), Filestore (NFS).
  • Databases: Cloud SQL (relational), Spanner (global SQL), Firestore (document), Bigtable (wide-column).
  • Data/ML: BigQuery (serverless SQL warehouse), Dataflow (stream/batch), Vertex AI.
  • Networking: VPC with subnet routing, Cloud Load Balancing (anycast GFE), Cloud DNS, Cloud NAT, interconnect.

GCP families and flagships.

Compute
Storage
Databases
Networking
Data & ML
Real world

In production

  • A start-up runs a web app on Cloud Run, Cloud SQL for data, Cloud Storage for uploads, and BigQuery for analytics.
  • A media platform runs distributed transcodes on Compute Engine spot, orchestrated by GKE.
  • A bank uses Spanner for globally consistent transactional data across US/EU regions.
Guidelines

Best practices

  • Learn the flagship of each family first; variants fill in later.
  • Use gcloud, not just the console; everything is reproducible via CLI or IaC.
  • Pick project structure and IAM early — reorganizing projects is expensive.
  • Prefer managed services (Cloud Run, Cloud SQL) over hand-running VMs where the workload allows.
Watch out

Common mistakes

  • Using Compute Engine for everything you could run on Cloud Run or Functions.
  • Forgetting Compute Engine still bills while stopped (disk costs).
  • Ignoring project-scoped IAM until resources are public.
  • Treating egress as free — cross-region data movement costs money.
Root cause analysis

Common issues

Error
Can't SSH to a Compute Engine VM
Root cause

Firewall rule, missing external IP or OS login mismatch

Fix

Check the network firewall-custom rules and VM's external IP/network tags.

Error
Cloud Run can't reach Cloud SQL
Root cause

Missing Cloud SQL connection or missing VPC connector

Fix

Add the Cloud SQL instance to the service's connections or route through the VPC connector.

Error
Bucket object 'forbidden'
Root cause

IAM missing, bucket policy conflict, or public ACL disabled

Fix

Grant storage.objectViewer at the bucket level and test with gsutil ls.

When things break

Debugging tips

  • Start at the family: if storage, check bucket ACL + IAM + CORS first.
  • Cloud Logging + monitoring charts cover most 'what changed' hunts.
  • For integration issues, read the exact error: 403 missingPermission vs 429 rateLimit have different fixes.
  • Use gcloud --format=json for tabular data that's easier to parse.
Interviews

Interview checkpoint

Test yourself

Quick quiz

0/2
1

Which GCP service is the managed, serverless SQL data warehouse?

2

GCP projects are the essential boundary for organizing resources and IAM.

Go deeper

Official documentation

GCP — Google Cloud product overview
docs
GCP — Compute Engine intro
docs