Stop 18 · Cloud & HostingTopics · 10/11

Google Cloud Platform (Overview)

In short

AWS isn't the only cloud. Google Cloud offers the same building blocks under different names, plus some favourites of its own like Cloud Run and BigQuery.

Google's cloud platform — a major alternative to AWS, with strong offerings in Kubernetes (GKE), serverless (Cloud Run) and data/ML services.

The purple blob compares two thali plates holding the same dishes arranged differently.

Think of it like this

Same thali, different restaurant

Every thali has dal, rice, roti and sabzi. A new restaurant just names them differently and plates them its own way. Learn one cloud well and the next one is mostly a new menu card: VMs, storage, functions and databases are all there.

Watch it happen

Same ideas, new names

AWS

EC2 for virtual machines
S3 for object storage
Lambda for functions

Google Cloud

Compute Engine for virtual machines
Cloud Storage for objects
Cloud Run for containers

The concepts carry over almost one to one. Learn the ideas once, then map the names.

The big ideas

What you need to know

Cloud Run

Runs any container that listens on PORT. It can scale to zero and bills for the time it's serving requests.

Cloud SQL and Firestore

Managed Postgres or MySQL with Cloud SQL, or the document database Firestore for mobile-style apps.

BigQuery

A serverless data warehouse. Run SQL over huge datasets without managing a single server.

Projects hold everything

Every resource lives in a project with its own billing and permissions. Make one per app and environment.

See it in code

Walk through the code

bash

1# Log in and pick a project2gcloud auth login3gcloud config set project my-shop-dev4 5# Build from source and deploy to Cloud Run in Mumbai6gcloud run deploy orders-api \7  --source . \8  --region asia-south1 \9  --allow-unauthenticated \10  --set-env-vars APP_ENV=staging11 12# Print the service URL13gcloud run services describe orders-api --region asia-south1 --format "value(status.url)"

1/3Log in and choose which project new resources go into.

--source . builds a container from your code (using your Dockerfile if there is one) and deploys it. The first run asks to enable the needed APIs.

Save yourself the pain

Common mistakes

Oops

Assuming services with the same purpose have the same limits and pricing.

Do this instead

Cloud Run, Lambda and Azure Functions differ in timeouts, concurrency and billing. Read the docs for the one you pick.

Oops

Putting dev, staging and prod in one project.

Do this instead

Use separate projects so a test script can't delete production, and each bill stays clear.

Your mission

Deploy to Cloud Run

~30 min

Quick check

Which Google Cloud service is closest to running your Dockerised Express API without managing VMs?

Go deeper

1Read GCP's rough equivalents to the AWS services you already know (EC2 → Compute Engine, S3 → Cloud Storage).
2Understand Cloud Run as GCP's popular option for containerized serverless deployments.
3Recognize most core cloud concepts transfer directly between AWS, GCP and Azure.