What is a Server?

Q: "What is the difference between a web server and an application server?"

"A web server serves static files, HTML, CSS, images, JavaScript. It handles HTTP requests and returns files. nginx and Apache are web servers. An application server runs your application's business logic, it processes requests, queries databases, applies rules, and generates responses dynamically. In practice, many deployments use nginx as a reverse proxy in front of a Node.js, Python, or Go application server. The two often run on the same machine."

Q: "What is serverless?"

"Serverless means you run code without managing server infrastructure. You write a function, deploy it to a provider (AWS Lambda, Cloudflare Workers, Vercel Functions), and the provider handles everything: starting, stopping, scaling, and billing per execution. You do not pay for idle time. Serverless is not literally without servers, there are still servers, but you never think about them. It is excellent for event-driven tasks and APIs with variable traffic."

Q: "What is a container, and how is it different from a virtual machine?"

"A virtual machine (VM) emulates a complete computer with its own OS, heavy, slow to start, isolated. A container shares the host OS kernel but packages an application with its dependencies in an isolated process, lightweight, starts in seconds, portable. Docker is the most common container tool. Running a containerised app on any server with Docker installed behaves identically, regardless of what else is on that server. This solved 'it works on my machine' problems."

A server is just a computer that runs continuously and waits for requests. When you load a website, a server somewhere in the world responds.

5 min read No prior knowledge needed

Recommended watch

What is a Server? Servers vs Desktops Explained

PowerCert Animated Videos

Watch on YouTube →

Quick Answer

A server is a computer, physically similar to your laptop, that is kept on 24/7, connected to the internet, and configured to respond to requests. When you visit a website, your browser sends a request to a server, which processes it and sends back a response. Every web app you use runs on servers.

The client-server model

Every interaction on the web is a conversation:

Client (your browser, phone, or app) sends a request: “Give me the homepage of example.com”
Server receives the request, processes it, and sends a response: the HTML of the homepage
Client receives and displays the result

This request-response pattern underpins everything: loading a web page, calling an API, sending a form, streaming a video. The details vary; the model is always the same. More: Client-server model, MDN

What makes a server different from your laptop?

Physically, almost nothing. A server is a computer with a CPU, RAM, and storage. What differs:

Always on, servers run 24/7. A web app cannot go offline at night.
No screen, servers are managed remotely via the terminal (SSH), not by sitting in front of them.
Optimised for throughput, servers handle thousands of simultaneous connections; your laptop handles one.
Located in data centres, dedicated buildings with backup power, cooling, redundant internet connections, and physical security.
Accessible by IP address, anyone with the right credentials can connect from anywhere in the world.

A laptop could technically be a server. In fact, your computer is a server when you run npm run dev, it listens on port 3000 and responds to requests from your own browser.

Types of servers

Type	Role	Examples
Web server	Serves static files over HTTP	nginx, Apache
Application server	Runs backend code, processes requests	Node.js, Python (Gunicorn), Go
Database server	Stores and retrieves structured data	PostgreSQL, MySQL, Redis
File/object storage server	Stores and serves files, images, videos	S3, Cloudflare R2, MinIO
Email server	Sends and receives email	Postfix, SendGrid
Cache server	Stores frequently accessed data in memory	Redis, Memcached
Reverse proxy	Sits in front of other servers, routes and load-balances	nginx, Cloudflare

In practice, one physical machine often runs several of these simultaneously. A small startup might run their entire stack on a single server. A large company might have hundreds of servers for each role.

Containers: solving “it works on my machine”

A classic problem: code works on your laptop but fails when deployed to the server. Different operating systems, different library versions, different configurations.

Docker solved this. You define your application and its dependencies in a Dockerfile. Docker packages everything into a container, a portable unit that runs identically on any machine with Docker installed. More: Docker getting started

dockerfile

FROM python:3.12
WORKDIR /app
COPY requirements.txt .
RUN pip install -r requirements.txt
COPY . .
CMD ["python", "app.py"]

This container will run exactly the same on your laptop, on AWS, on Google Cloud, or in any data centre. No “works on my machine” anymore.

Kubernetes orchestrates containers at scale, deciding which server runs which containers, restarting failed containers, scaling up under load. Most large deployments use it. More: Kubernetes

Serverless: code without infrastructure

Serverless platforms (AWS Lambda, Cloudflare Workers, Vercel Functions, Netlify Functions) let you deploy individual functions rather than whole applications. You write a function, push it, and the platform:

Runs it on request
Scales automatically from zero to millions of executions
Bills you per invocation, not per hour

For a simple API endpoint or a webhook handler, serverless is often the simplest and cheapest option. For long-running processes or stateful applications, it is less suitable.

Where to host your first project

Platform	What it is	Best for
Vercel	Serverless deployment with Git integration	Next.js, React apps, APIs
Railway	Simple container hosting	Any backend, databases
Render	PaaS with free tier	APIs, web services, cron jobs
Fly.io	Container deployment close to users	Low-latency global apps
AWS / GCP / Azure	Full cloud platforms	Production apps, complex infrastructure
DigitalOcean	VPS and managed services	More control than PaaS, less than raw AWS

For a prototype or demo: start with Vercel, Railway, or Render. You push code from GitHub and it deploys in minutes. No server management required.

The request lifecycle

What actually happens when someone visits your web app:

Browser resolves the domain to an IP address via DNS
TCP connection established with the server
Browser sends HTTP request
nginx (or another reverse proxy) receives it, passes to your app server
Your app server runs your code: reads the request, queries the database, applies logic
Database server returns data
App server formats the response (JSON, HTML)
nginx sends the response back to the browser
Browser renders the result

This happens in under 100ms for well-optimised apps.

Request lifecycle: browser to database and back

Client Browser Sends HTTP GET to the domain. DNS resolves it to an IP.

→

Edge / Proxy nginx / CDN Terminates TLS, serves static files, routes dynamic requests to the app.

→

App server Your code Processes request, applies business logic, queries database.

→

Data layer DB + Cache PostgreSQL for persistent data; Redis serves cached results in microseconds.

A dark industrial floor with red neon grid seams laid in precise layers: infrastructure built in sections, each zone isolated and addressable. — A production server stack. Each layer handles a specific concern: edge caching, load distribution, application logic, data persistence. This is what deploying to the cloud actually means. The grid is already there. You configure how traffic moves through it.

What’s next

Next: What is the Cloud? , what “the cloud” actually means and why everyone moved there.

Official documentation: What is a web server?, MDN

Frequently asked questions

What is the difference between a web server and an application server?

A web server serves static files, HTML, CSS, images, JavaScript. It handles HTTP requests and returns files. nginx and Apache are web servers. An application server runs your application’s business logic, it processes requests, queries databases, applies rules, and generates responses dynamically. In practice, many deployments use nginx as a reverse proxy in front of a Node.js, Python, or Go application server. The two often run on the same machine.

What is serverless?

Serverless means you run code without managing server infrastructure. You write a function, deploy it to a provider (AWS Lambda, Cloudflare Workers, Vercel Functions), and the provider handles everything: starting, stopping, scaling, and billing per execution. You do not pay for idle time. Serverless is not literally without servers, there are still servers, but you never think about them. It is excellent for event-driven tasks and APIs with variable traffic.

What is a container, and how is it different from a virtual machine?

A virtual machine (VM) emulates a complete computer with its own OS, heavy, slow to start, isolated. A container shares the host OS kernel but packages an application with its dependencies in an isolated process, lightweight, starts in seconds, portable. Docker is the most common container tool. Running a containerised app on any server with Docker installed behaves identically, regardless of what else is on that server. This solved ‘it works on my machine’ problems.