CH. 04 The cloud · a planet-sized computer

What is the cloud, actually?

Not a mystical place. It's millions of computers in buildings around the planet — always on, always connected, rented out by the second. When you want the whole world to reach your app, this is where it goes.

See inside the cloud Put my app there

Hours a day: 24
Billed by: The second
Reachable from: Anywhere

Fig. 01 — The internet, from orbitDrag to spin

Cloud regions (illustrative)You · approx., from your time zoneRequests in flight

§ 01 The short answer

The cloud, in one paragraph

Quick answer

The cloud is computing you rent instead of own: always-on computers in data centers, with fast connections and public internet addresses, run by companies like Amazon, Google, Microsoft and DigitalOcean. You pay for what you use — often by the second.

“Deploying to the cloud” means copying your app from your laptop (your localhost) onto one of those machines, so anyone can reach it, anytime, even with your laptop closed.

Origin Why “cloud”?

Fig. 02 — How engineers drew the networkLive

On network diagrams, engineers drew the parts they didn't control — the telephone network, later the internet — as a puffy cloud. When companies began renting out computers you reach over the internet, the name followed. It went mainstream in 2006, the year Amazon launched EC2 and started renting servers by the hour.

§ 02 Anatomy

Zoom into the cloud, detail by detail

From a whole metro area down to the slice your app gets. Each drawing magnifies the highlighted part of the one before it.

Detail A — Region→ B

Region. A cluster of data centers in one metro area — Northern Virginia, Frankfurt, Tokyo. Big clouds run dozens of regions; you pick the one closest to your users.

Detail B — Availability zone→ C

Availability zone. One or more data centers with independent power, cooling and networking, kilometers away from the others — so one fire or flood can't take a whole region down.

Detail C — Data center→ D

Data center. A warehouse of server racks in rows, with alternating hot and cold aisles, redundant power, backup generators and industrial cooling. A large one holds tens of thousands of servers.

Detail D — Rack→ E

Rack. A standard rack is 42 “units” tall — each unit 1.75 inches (44.45 mm) — packed with servers, power strips and a network switch at the top.

Detail E — Server→ F

Server. Dozens to hundreds of CPU cores and hundreds of gigabytes of memory, carved by a hypervisor into many isolated virtual machines for many customers.

Detail F — Your appLive on :8080

Your app. Your container gets a slice — maybe half a CPU core and a gigabyte of memory — and one open port that the internet can reach through a public address.

§ 03 In motion

What happens when someone opens your app

Ten messages, five machines, a few thousand kilometers — and it's done in about the time it takes to blink.

Fig. 03 — One page load, step by stept = 0 ms ·

Illustrative timings for a visitor about 2,000 km from the data center. The first steps cost whole round trips across the internet; inside the data center, hops take about a millisecond — which is why distance matters so much.

§ 04 Shared responsibility

IaaS, PaaS, SaaS: who manages what?

Every kind of cloud is the same stack of layers. The only question is where your job ends and the provider's begins.

Your own serversOn-premises

EC2, Droplets, Compute EngineIaaS · VMs

Cloud Run, ECS, CloudishContainers

Vercel, Heroku, LambdaPaaS · Serverless

Gmail, Notion, FigmaSaaS

Your data

You

App code

You

Provider

Runtime

You

You · in image

Provider

Operating system

You

Provider

Virtualization

You

Provider

Servers & storage

You

Provider

Networking

You

Provider

Building, power, cooling

You

Provider

You manage itThe provider manages it

§ 05 The speed of light is a budget

Why distance matters

Light in fiber travels about 200 km per millisecond — two-thirds of its speed in a vacuum. That sounds instant until you do the math: a round trip from New York to London covers about 11,000 km, so it can never take less than ~56 ms, no matter how fast the servers are.

That's why clouds have regions on every continent, and why you put your app near your users. And it's why localhost feels impossibly fast: its round trip has no distance in it at all.

The orange bar is a live measurement: your browser's round trip to the server hosting this page, just now.

Fig. 04 — Typical round-trip timesLog scale · each line = 10×

Loopback · localhostyour computer → itself

0.05 ms

Same Wi-Filaptop → phone

3 ms

Same cityyou → a nearby data center

10 ms

Across a continentNew York ↔ Los Angeles

65 ms

Across the AtlanticNew York ↔ London

75 ms

Around the worldLondon ↔ Sydney

260 ms

Geostationary satelliteup 35,786 km and back

600 ms

You → this websitemeasured live

measuring…

Show as a table

Trip	Typical round trip
Loopback (localhost)	0.05 ms
Same Wi-Fi network	3 ms
Same city	10 ms
New York ↔ Los Angeles	65 ms
New York ↔ London	75 ms
London ↔ Sydney	260 ms
Geostationary satellite	600 ms
You → this website (live)	—

Under the sea

1.4M+ km

of fiber-optic submarine cable carry almost all intercontinental internet traffic.

Light in glass

≈200 km/ms

The hard speed limit for every request your app will ever answer.

A standard rack

42U

Units of 1.75 inches, stacked floor to ceiling with servers.

The modern cloud began

2006

When Amazon launched S3 and EC2 and began renting servers by the hour.

§ 06 Side by side

Your laptop vs. a cloud server

Localhost isn't broken — it's just the wrong kind of home for anything that needs to outlive your laptop being open.

Feature	Your laptop (localhost)	A cloud server
Stays on 24/7	No — it sleeps, closes, travels	Yes — by design
Reachable from anywhere	No — only from itself (or your Wi-Fi)	Yes — it has a public IP address
A real URL like my-app.com	Not practically	Yes — point DNS at it
HTTPS for visitors	No	Yes — usually automatic
Handles many visitors at once	Barely — it's also your laptop	Yes — add capacity as you grow
Who maintains the hardware	You	The provider
Cost	Free — you already own it	Pay for what you use; small apps can cost cents a day

§ 07 The landscape

Kinds of clouds

All of them rent you computers in data centers. They differ in how much they hide from you.

Hyperscalers

AWS · Google Cloud · Azure

Hundreds of services in dozens of regions. Maximum power and flexibility — and a learning curve to match.

Developer clouds

DigitalOcean · Hetzner · Akamai

Simple virtual servers and managed databases with predictable pricing. Great when you want a real server without the sprawl.

App platforms

Vercel · Netlify · Render · Railway · Fly.io

Push your code; they handle servers, scaling and HTTPS. The default for many web apps.

Backend as a service

Supabase · Firebase

A managed database, logins and file storage you plug into, instead of running your own backend.

Agent-first

Cloudish.ai

No dashboard at all: an API built for AI coding agents to deploy containers end-to-end, from one sentence. This site runs on it.

Compare them

Which one should I pick? →

Commands, trade-offs and a straight answer for your situation in the deploy guide.

§ 08 FAQ

Cloud questions, answered

What is the cloud, in simple terms?

Computers in data centers that you rent over the internet instead of buying. They stay on all the time, have fast connections and public addresses, and you pay only for what you use.

Where is the cloud, physically?

In data centers: warehouse-sized buildings full of servers, grouped into regions near big cities on every inhabited continent. When you deploy, you usually choose a region close to your users.

Is the cloud just someone else's computer?

Yes — millions of them, with redundant power, cooling, networking and security teams you'd never build yourself, available in seconds and billed by usage. “Someone else's computer” is exactly the point.

What's the difference between IaaS, PaaS and SaaS?

They differ in how much you manage. With IaaS you rent virtual machines and manage the operating system and everything above it. With PaaS you hand over your code and the platform runs it. With SaaS you just use finished software, like Gmail.

Why does my app need the cloud if it works on localhost?

Localhost only works on your own computer and only while it's on. A cloud server has a public address and never sleeps, so anyone can use your app from anywhere, at any time.

Is the cloud expensive?

It doesn't have to be. Many platforms have free tiers, bill by the second, and scale idle apps to zero, so a small personal project can cost cents a day or nothing at all. Costs grow with traffic and resources.

Next · Chapter 05Deploy your app — in one sentenceAsk an AI agent, or pick a platform. From localhost to a live URL in minutes.