The Name Is Not the House

You buy a domain. You pay ₦12,000 for myproject.com. You feel like a founder. You open your browser, type the name, and press enter — and get an error page. Nothing is there. Your domain exists, but your website doesn’t.

Because you just bought an address, not a house.

This is the single most common confusion in web development, and it costs people real money every day. They buy domains thinking they’re buying websites. They buy hosting thinking they’re buying domains. They pay two different companies and don’t understand why. They see “the cloud” on a pricing page and have no idea what it actually means.

Today, we separate all of it. Domains. Hosting. The cloud. The ladder of options. And a proper backstory about how this industry got so weird.

The name is not the house

Let me put the whole thing in one analogy, because everything else in this lesson builds on it.

A domain name is like a plot of land with an address. A hosting provider is the builder who constructs the house on that land. You can buy land in Lekki and build a house on it. You can rent someone else’s house while owning land elsewhere. You can even sell the house and keep the land. But they are two separate things, and the fact that they’re usually sold together — by companies that want you to buy both — hides that distinction.

Domain name — A human-readable identifier for a location on the internet, structured as a hierarchy of labels separated by dots, registered through an accredited registrar.

In plain terms: the address you own and renew. myproject.com, readunderstory.ink, google.com — those are domain names.

Humans can’t remember IP addresses, and DNS needs a stable identifier to map to those IPs. A domain name is that identifier — the human-facing name on the envelope, not the machine-facing address underneath it.

The name on a plot of land in a government registry. Everyone can look it up, everyone can reference it, but only you’re the registered owner.

Except land is physical and the ownership rules are national. Domain names are virtual and their ownership is managed by a global non-profit (ICANN) through a network of private companies. It’s the same idea, but the “land registry” is global, private, and moves much faster.

What a domain name actually is

A domain name is a structured string. Let’s break down readunderstory.ink:

readunderstory.ink
       ↑       ↑
       │       └── TLD (top-level domain)
       └────────── second-level domain

Read right to left — the way DNS actually reads it.

The TLD (top-level domain) is the last part. .com, .org, .ink, .ng, .io — each one is managed by a different organisation. .com is managed by Verisign. .ng is managed by the Nigerian Internet Registration Association. .ink is managed by a company called Top Level Design.

The second-level domain is the part you actually get to choose. That’s the “readunderstory” part. When you “buy a domain,” you’re registering the second-level name under a specific TLD.

The subdomain is the optional part that comes before the second-level name. You don’t buy these — you create them. www.readunderstory.ink, api.readunderstory.ink, blog.readunderstory.ink — those are subdomains that you control entirely.

TLD — Top-Level Domain — The rightmost label in a domain name, such as .com, .org, or .ng. Managed by an organisation designated by ICANN.

In plain terms: the ending of the domain — the branch of the tree that says what kind of site this is.

TLDs let the domain system partition itself. .com names are managed by one organisation, .ng names by another. This lets each TLD have its own rules, registration process, and pricing.

The surname in a person’s full name. Everyone in the “Adeyemi” family shares a name, but each individual has their own first name. The TLD is the surname — a shared category with shared rules.

Except surnames are inherited. TLDs are chosen, purchased, and policed by an organisation. You can’t just invent a new TLD — you have to apply to ICANN, pay hundreds of thousands of dollars, and pass a technical evaluation.

How you actually get one

You don’t buy a domain from the internet. You register it through a registrar — a company accredited to sell domains.

Registrar — A company accredited by ICANN to register domain names on behalf of end users and manage their lifecycle.

In plain terms: the shop that sells you the domain and renews it every year.

ICANN doesn’t sell domains to individuals — it accredits hundreds of companies worldwide to do so, which keeps the market competitive and distributes the technical and commercial work.

The car dealership. The manufacturer (ICANN) builds the product, but you buy it from a dealer (the registrar).

Except dealerships can lose accreditation. ICANN can revoke a registrar’s right to sell domains if they break the rules, which means your domain could be frozen if your registrar gets shut down. The way out is called a “transfer” — moving your domain to a different registrar, which you’re entitled to do.

Common registrars: Namecheap, Cloudflare Registrar (sells at cost), Porkbun, GoDaddy (the expensive one people warn you about), plus Whogohost and QServers in Nigeria.

The registry is the entity that actually owns the database of all domains under a TLD — for example, Verisign owns the .com registry. Registrars are customers of registries. When you register a domain through Namecheap, Namecheap is registering it with Verisign on your behalf.

Registry — The organisation that maintains the authoritative database of all domains registered under a given top-level domain.

In plain terms: the actual ledger. The registrar is the shop; the registry is the warehouse the shop orders from.

You renew your domain annually. Miss the renewal, and the domain lapses — then it goes into a “redemption period” (30 days), then a “pending delete” period (5 days), then it becomes available again. If your domain lapses and someone else grabs it, it’s theirs. Not yours.

This is why serious businesses register their domain for multiple years at a time. A missed renewal is one of the most expensive mistakes in tech — it can end companies.

ICANN — The Internet Corporation for Assigned Names and Numbers, the non-profit organisation responsible for coordinating the domain name system and internet numbering policy globally.

In plain terms: the referee. It doesn’t sell domains or host websites — it sets the rules both sides operate under.

WHOIS — A query protocol and public directory that returns registration information for a domain name.

In plain terms: the phone book entry for a domain. Who’s it registered to, when does it expire, who’s the nameserver.

WHOIS is one of the oldest parts of the internet, and it’s still how people check whether a domain is available, who owns one, or when it expires. Privacy services that hide your details behind WHOIS exist for a reason — historically it exposed personal contact information publicly.

What hosting actually is

Now the house.

Hosting is the server, storage, and infrastructure where your website actually runs. It’s the physical computer (or virtual one, or container, or serverless function) that responds to requests for your site.

Hosting — The service of providing server infrastructure — compute, storage, and network — on which websites and applications run.

In plain terms: the “where” of your website. The computer that lives at the address your domain points to.

A domain name is just a name. Something has to be there, answering requests. That “something” is hosting.

The house on the plot of land. The address is useful, but only because there’s a building there to visit.

Except a house is a fixed thing. Hosting is a service — your site might be running on a shared machine, or a virtual slice of a machine, or a container that spins up on demand. The word “hosting” covers all of these.

The ladder of hosting options

Hosting comes in several tiers, from cheap and simple to expensive and powerful. Every developer should know the ladder, because choosing the wrong rung costs money or limits your growth.

Shared hosting — you share a single server with hundreds of other websites. The hosting provider manages everything: the operating system, the web server, the database. You just upload files. Cheap, usually ₦3,000–₦15,000/month. Slow if your neighbours use too many resources. Fine for static sites and small WordPress blogs.

Shared hosting — A hosting model in which multiple websites share the same physical server and its resources.

In plain terms: a room in a shared apartment. Cheap, convenient, and your neighbours’ noise is your problem.

VPS — Virtual Private Server — you get a virtual machine that feels like your own dedicated server, but it’s running on shared hardware. You get root access, you install what you want, you manage updates. Around ₦10,000–₦80,000/month depending on specs. This is the standard for developers learning to deploy. DigitalOcean, Linode, Vultr, and Hetzner are the big names.

VPS — Virtual Private Server — A virtualised server instance that provides dedicated compute resources on shared physical hardware, typically with full root access.

In plain terms: your own server, but virtual. It feels like a full computer because it is one — logically.

Dedicated server — a whole physical machine, just for you. No sharing, no neighbours. Expensive (hundreds to thousands of dollars a month). Only worth it for high-traffic production systems.

Dedicated server — A physical server leased exclusively to a single customer.

In plain terms: the building itself, with no neighbours. Maximum control, maximum price.

PaaS — Platform as a Service — the host manages the server for you. You push code, they run it. You don’t deal with Nginx, SSL, or deployments. Vercel, Netlify, Render, Railway, Fly.io, and Heroku are the big names. Great for getting products to production quickly; more expensive at scale.

PaaS — Platform as a Service — A hosting model where the provider manages the runtime, operating system, and infrastructure, so the developer only provides application code.

In plain terms: a serviced apartment. The building is maintained for you; you just move in and live.

IaaS — Infrastructure as a Service — the provider gives you virtualised compute, storage, and networking, and you assemble it. AWS EC2, Google Compute Engine, Azure VMs. You have more control than PaaS and more responsibility.

IaaS — Infrastructure as a Service — A hosting model where the provider offers raw virtualised compute, storage, and networking, which the developer configures and manages.

In plain terms: an empty flat with the bills in your name. The walls are there; the wiring, the furniture, and the internet connection are your problem.

SaaS — Software as a Service — finished software you use over the internet. Gmail, Salesforce, Figma, Notion. You don’t host any of it.

SaaS — Software as a Service — A software distribution model where applications are hosted by a provider and accessed by users over the internet.

In plain terms: you rent the finished product, not the machinery behind it.

Serverless — you deploy individual functions, and the platform runs them on demand. You only pay for the milliseconds of compute you actually use. AWS Lambda, Google Cloud Functions, Cloudflare Workers, Vercel Functions. Beautiful for small workloads; tricky to reason about at scale.

Serverless — A cloud execution model in which the provider dynamically allocates compute resources on demand, and the customer is billed only for actual usage.

In plain terms: hiring a taxi for one trip instead of keeping a car parked all day.

Static hosting — for sites that don’t need a backend. Just HTML, CSS, and JS files served from a CDN. GitHub Pages, Cloudflare Pages, Netlify, Vercel. Free or nearly free.

Static hosting — A hosting model that serves pre-built HTML, CSS, and JavaScript files directly, without a backend runtime.

In plain terms: a shop with the goods already on the shelves. No kitchen, no machinery — just the things customers came for.

The tier you’ll actually use

For your own projects and most freelance work, you’ll live in one of three places:

• Vercel or Netlify — for frontend apps and small full-stack projects. Free tier is generous. Deploy in seconds. You’ll learn this first because it’s the easiest.
• A VPS — for backends that need persistent state, custom software, or that outgrow the free tiers. Hetzner and DigitalOcean are the standard choices.
• A “boring” shared host — for a WordPress site for a client who wants nothing fancy. This is still how a lot of real-world client work happens.

You don’t need to know all of these on day one. You need to know they exist, and roughly what each one costs in complexity and money. That’s how you make good decisions instead of just following a YouTube tutorial.

The cloud, demystified

Now the phrase that pays.

“The cloud” is not a place. It’s not a technology. It’s a marketing term for “someone else’s servers, that you rent instead of own, accessed over the internet.”

The cloud — A colloquial term for computing resources — servers, storage, databases — provided over the internet by a third party, rather than owned and operated on-premises.

In plain terms: renting someone else’s computers instead of buying your own.

Before the cloud, every company that wanted to run software had to buy servers, rack them in a data centre, keep them cool, replace the failing ones, hire people to maintain them. This was slow and expensive. The cloud turned that into an API call.

Buying a generator and hiring a mechanic versus plugging into the grid. Same electricity, hugely different overhead.

Except the cloud isn’t guaranteed like the grid — it’s more like a very reliable private generator you rent from a company. When it goes down, you can’t call the landlord and demand power. You can only wait — and hope you built your system to survive the outage.

How the cloud actually works

Physically, the cloud is data centres — massive buildings full of servers. The biggest operators are Amazon (AWS), Microsoft (Azure), and Google (GCP). Between them, they own the vast majority of cloud infrastructure in the world.

The magic is virtualisation. A single physical server, running a piece of software called a hypervisor, can be split into dozens or hundreds of virtual machines. Each VM behaves like its own computer — its own OS, its own network, its own storage. But it’s running on shared hardware.

Virtualisation — The process of creating a virtual version of a computing resource — a server, storage, or network — that behaves like the real thing while sharing underlying physical hardware.

In plain terms: one computer pretending to be many.

Servers were expensive and often sat underused — most of the time, most servers run at 5–15% capacity. Virtualisation lets customers share that unused capacity, making hardware dramatically cheaper to operate and more efficient to use.

A landlord splitting a six-bedroom house into six flats. Same building, six tenants, each with their own key.

Except flats share a foundation, a roof, and a plumbing system. If something structural breaks, all six flats feel it. Virtual machines have the same problem — a bad neighbour (a “noisy neighbour” in cloud terms) can degrade performance for everyone on that physical host. That’s why cloud providers sell “dedicated” tiers.

Hypervisor — Software that creates and manages virtual machines on a physical host.

In plain terms: the landlord’s clerk. It carves the building into flats and hands out the keys.

Regions and availability zones

Two more pieces of cloud vocabulary you’ll hear constantly.

Region — A geographic area in which a cloud provider operates a cluster of data centres, such as us-east-1 (Virginia) or af-south-1 (Cape Town).

In plain terms: which part of the world your server runs in.

Latency. Physics. Every additional thousand kilometres between your server and your user costs real milliseconds of round-trip time. Choosing the right region is a performance decision.

A courier’s local depot. The closer your depot is to the delivery address, the faster the package arrives.

Except a depot serves a fixed town. A cloud region serves a continent, and most cloud providers have very few regions in Africa. This is why developers building for Nigerian users often see higher latencies than developers in the US — a real, ongoing concern we’ll touch on in Module 13.

Availability zone (AZ) — An isolated data centre (or group of data centres) within a region, engineered so that a failure in one zone does not affect others.

In plain terms: a spare data centre next door, so if one burns down, the other keeps running.

Cloud providers promise “high availability” — reliability numbers like 99.99%. That promise only holds if your workload is spread across multiple zones. A single-zone deployment is not highly available, even if the cloud provider is.

A large company with two factories in the same city. If one catches fire, the other keeps producing.

Except the two factories usually share roads, power, and water. Availability zones are engineered with separate power, cooling, and network links — so a single failure really can’t take both down. This is expensive, which is why most providers only offer it in a handful of premium regions.

How the name finds the house

You now know the two halves. Let’s connect them.

When you buy a domain and buy hosting, how does one find the other? This is where Day 7 comes back.

Step 1: Your registrar asks you which nameservers to use. By default, they point at the registrar’s own nameservers. Most people change this to point at a DNS provider like Cloudflare.

Step 2: You add DNS records. At your DNS provider, you add records that tell the world where your site actually lives:

• For a static site (Vercel, Netlify), you add a CNAME or A record pointing at the host.
• For a VPS, you add an A record pointing at your server’s IP address.
• For a subdomain like api.myproject.com, you add its own record.

Step 3: DNS propagates. Once the records are set, the whole internet starts resolving your domain to your hosting. This can be instant (if no cache) or take hours (if cached with a long TTL).

Step 4: A request arrives. Someone types your domain. DNS resolves it to your host’s IP. The host’s server receives the request, checks which domain the request was for (from the Host header), and serves the right site.

Step 5: TLS is provisioned. Most modern hosts — Vercel, Netlify, Cloudflare — automatically issue an SSL certificate for your domain once DNS is set up. Within a minute or two, HTTPS starts working. This used to be a multi-day manual process; now it’s automatic. This is one of the biggest quality-of-life changes in the last decade.

The whole setup takes maybe 10 minutes if you know what you’re doing. It takes three frustrated hours if you don’t. Today is about making you the 10-minute version.

Why prices vary so wildly

You’ll see hosting that costs ₦1,500/month and hosting that costs ₦2,000,000/month. Why?

Shared hosting is cheap because you’re sharing almost everything — CPU, memory, disk, and the maintenance cost — with hundreds of other customers. It’s the budget option, and it shows: your site’s speed depends partly on your neighbours’ behaviour.

VPS is mid-priced because you get dedicated resources, but the physical hardware is still shared. Providers like Hetzner have brought the price down to under $5/month for a small VPS.

PaaS is more expensive per unit of compute because the platform abstracts away all the infrastructure work. You pay for convenience, deployment automation, and reliability. Vercel’s free tier is generous; Vercel’s paid tier is considerably more expensive than running the same workload on a VPS. You’re paying for time, not raw compute.

Dedicated servers and enterprise cloud are expensive because you’re buying exclusive hardware or paying for features large companies need: compliance, guaranteed SLAs, private networking, dedicated support.

The real question isn’t “which is cheapest” — it’s “which is cheapest for this project at this stage.” A weekend side project belongs on a free tier. A serious SaaS with paying customers belongs on a paid PaaS or a managed VPS. A high-traffic production system belongs on cloud infrastructure with a proper reliability budget. Match the tool to the stage. That’s the entire skill.

Backstory: how this industry got so weird

You asked for backstory. This one’s a proper story.

1985 — The first domain. symbolics.com is registered on March 15, 1985. It’s still registered today. It’s the oldest domain name on the internet.

1985–1992 — Domain registration is free. You email a person at Stanford Research Institute, and they add your domain to the list. No cost, no registrar, no paperwork. This is how the early internet worked.

1993 — Domain registration becomes commercial. The US government contracts with Network Solutions to manage the whole .com, .net, and .org registry. Network Solutions starts charging for domains, and for several years, it has a de facto monopoly. Prices are high, service is slow, and nobody can compete.

1998 — ICANN is founded. The US Department of Commerce creates the Internet Corporation for Assigned Names and Numbers to take over DNS governance. The goal: break Network Solutions’ monopoly, open up the registrar market, and internationalise the management of the internet. This is the moment “registrar” becomes an actual business category — dozens of companies get accredited, and prices start falling.

2000 — The dot-com bubble burst. A lot of domains that were registered by speculators trying to flip them get dropped. This is when the secondary market for domain names starts getting serious — thousands of dollars paid for a single domain.

2002 — AWS launches. Amazon’s cloud computing platform opens. Initially it’s just S3 (storage) and SQS (queue). EC2 (virtual servers) comes in 2006. The world’s first “you can rent a computer by the hour” service. This is the moment “the cloud” starts becoming a real thing that people outside of enterprise IT can use.

2006 — EC2 launches. Virtual machines on demand. Developers can now spin up servers in seconds instead of weeks. Startups can build products without buying physical hardware. This is arguably the single most important moment in modern web infrastructure.

2008 — Heroku launches. The first widely-used PaaS. You git push and your app is live. No servers, no config, no SSH. Developers fall in love. The idea that “deploying = pushing code” starts to become normal, and eventually leaks into the modern standard (Vercel, Netlify).

2014 — Serverless enters the mainstream. AWS Lambda launches. You can now run code without provisioning any server. The unit of deployment becomes a single function. The pricing model becomes “pay for milliseconds of execution.” This changes the economics of small workloads dramatically.

2014–2016 — Static hosting gets radical. Netlify launches in 2014, Vercel in 2015 (as ZEIT). Both build a model where you git push a frontend and get global CDN hosting, automatic SSL, and zero config. Within a few years, deploying a modern frontend becomes genuinely a two-minute job.

2020s — Consolidation and verticalisation. Cloud providers expand into more and more verticals. Vercel adds a backend. AWS launches a dozen services per year. There’s a whole category of “we do one thing really well” tools — Cloudflare for DNS and security, Render for PaaS, Railway for quick deployments, Supabase and Firebase for the backend-as-a-service tier. The landscape is crowded and fascinating.

The takeaway: the “one registrar, one host, one provider” model that dominated the 90s and 2000s is gone. Today, most serious web projects use five to ten different services, all stitched together by DNS, APIs, and CI/CD. Understanding each layer — domain, DNS, hosting, deployment, database, storage — is now the actual skill. That’s why today’s lesson matters, even though it doesn’t involve writing a single line of code.

What you’d say in the room

If an interviewer asks “what’s the difference between a domain and hosting?” — here’s the answer:

A domain name is a human-readable identifier that resolves, via DNS, to an IP address. It’s a naming and routing layer, not a machine — you rent it from a registrar and renew it annually. Hosting is the actual computing infrastructure that sits at that address and answers the requests. They’re bought from different companies, they fail for different reasons, and you can point a domain at any host or move a host behind a domain without touching the other half.

If an interviewer asks “what’s the difference between PaaS and serverless?” — here’s the answer:

PaaS gives you a complete application runtime that you deploy to as a unit — your app runs continuously, and you’re billed for the instance it lives on. Serverless gives you individual functions that are invoked per event and scaled to zero between calls, billed per invocation and per millisecond of execution. The practical difference is billing granularity and cold starts: PaaS is simpler and better for long-running or stateful apps; serverless is cheaper for spiky, event-driven work, at the cost of execution latency and a harder debugging story.

Say it like a professional

The definitions from today, in one block. Say each out loud. If you stumble, go back.

• Domain name — A human-readable identifier for a location on the internet, structured as a hierarchy of labels separated by dots, registered through an accredited registrar.
• Registrar — A company accredited by ICANN to register domain names on behalf of end users and manage their lifecycle.
• Registry — The organisation that maintains the authoritative database of all domains registered under a given top-level domain.
• TLD — Top-Level Domain — The rightmost label in a domain name, such as .com, .org, or .ng. Managed by an organisation designated by ICANN.
• ICANN — The Internet Corporation for Assigned Names and Numbers, the non-profit organisation responsible for coordinating the domain name system and internet numbering policy globally.
• WHOIS — A query protocol and public directory that returns registration information for a domain name.
• Hosting — The service of providing server infrastructure — compute, storage, and network — on which websites and applications run.
• Shared hosting — A hosting model in which multiple websites share the same physical server and its resources.
• VPS — Virtual Private Server — A virtualised server instance that provides dedicated compute resources on shared physical hardware, typically with full root access.
• Dedicated server — A physical server leased exclusively to a single customer.
• PaaS — Platform as a Service — A hosting model where the provider manages the runtime, operating system, and infrastructure, so the developer only provides application code.
• IaaS — Infrastructure as a Service — A hosting model where the provider offers raw virtualised compute, storage, and networking, which the developer configures and manages.
• SaaS — Software as a Service — A software distribution model where applications are hosted by a provider and accessed by users over the internet.
• Serverless — A cloud execution model in which the provider dynamically allocates compute resources on demand, and the customer is billed only for actual usage.
• Static hosting — A hosting model that serves pre-built HTML, CSS, and JavaScript files directly, without a backend runtime.
• Virtualisation — The process of creating a virtual version of a computing resource that behaves like the real thing while sharing underlying physical hardware.
• Hypervisor — Software that creates and manages virtual machines on a physical host.
• Region — A geographic area in which a cloud provider operates a cluster of data centres.
• Availability zone — An isolated data centre within a region, engineered so that failure in one zone does not affect others.
• The cloud — A colloquial term for computing resources provided over the internet by a third party, rather than owned and operated on-premises.

Close the tab

Reproduce these from memory. Then reopen and check.

1. What’s the difference between a domain name and hosting?
2. What is a TLD? What is a second-level domain? What is a subdomain?
3. What is a registrar, and what role does ICANN play?
4. What is a registry, and how does it relate to a registrar?
5. What happens if you forget to renew your domain?
6. Name five hosting models and give a one-line description of each.
7. What is virtualisation, and what is a hypervisor?
8. What is the difference between a region and an availability zone?
9. What is “the cloud,” physically speaking?
10. Why might the same workload cost ₦5,000/month on a VPS and ₦150,000/month on a PaaS?
11. How do you connect a domain to a host? Which DNS records are involved?
12. Why did the cloud industry explode after 2006?

If you can answer all twelve without peeking, Day 8 is done. If not, read those sections again.

Tomorrow

You now know how a request travels — from URL to DNS to host to server. You know what verbs it can carry, what status codes come back, and where the machine lives that answers.

But there’s still a piece of the request we’ve been glossing over: what else travels with it.

The headers. The cookies. The sessions. The little pieces of context and memory that turn an anonymous request into an authenticated, personalised, stateful experience — even though HTTP itself remembers nothing.

Tomorrow, we meet headers, cookies, and sessions. You’ll learn how a website remembers you after you log in, why cookies aren’t just tracking devices, what “session” really means on the server side, and what to do when a cookie is silently breaking everything.