Cloud Computing

⏱ 8 min✏️ Quiz at the end

There Is No Cloud…

…goes the joke, only other people's computers. And the joke is accurate. "The cloud" is not vapour: it is data centres — warehouse-sized buildings packed with racks of servers, humming behind security fences, connected to the internet by pipes of staggering capacity. When your photo "uploads to the cloud", it travels as packets to one of those buildings and lands on real storage devices owned by a provider.

Cloud computing means using computers you reach rather than computers you own — storage, processing power, and software, delivered over the internet, rented as needed.

You Already Live There

If you use webmail, online documents, photo backup, streaming, or a game whose save-file follows you between devices — you are a cloud user. Notice what those share: your stuff is not on your device. The device is a window; the data and much of the work live in the data centre. Which explains the everyday magic:

  • Log in from any device and your world is there
  • Drop the laptop in a lake; the homework survives
  • Two people edit one document simultaneously — because there is really only one document, in the data centre, with both of you looking through windows at it

The Business Revolution: Renting vs Owning

For companies, the cloud changed the economics of computing itself. The old way: buy your own servers — guessing years ahead how many you'll need. Guess high and expensive machines idle; guess low and your site collapses on your best day. Plus a server room, cooling, and staff.

The cloud way: rent, by the hour, exactly what you use — and change your mind whenever. Giants like Amazon, Microsoft, and Google run world-spanning data centres and lease slices of them to everyone else. Startups launch with no hardware at all; the app on your phone almost certainly runs its backend on rented cloud servers.

The killer feature is elasticity. Your game goes viral tonight and needs 100× the capacity? Rent it within minutes; release it when the rush fades. A ticket site expecting a famous on-sale morning scales to thousands of servers for six hours. This on-demand scaling — impossible in the owned-hardware world — is arguably the enabling technology of the modern app economy. (Handling that grade of concurrent demand is the database reliability story at cloud scale.)

Inside the Building

Data centre engineering is its own marvel:

  • Redundancy everywhere — your data is replicated: copied across machines, often across sites, so a failed drive (and drives fail daily at this scale — remember, every storage device eventually fails) loses nothing. The provider is running 3-2-1 backup thinking, industrialised.
  • Heat is the enemy — hundreds of thousands of processors make data centres colossal electricity consumers, and cooling them is half the battle. Where that electricity comes from is a genuine environmental question; big providers now build wind and solar farms and site centres in cold climates partly for free cooling.
  • Security, physical and digital — fences and guards outside; encryption and access control within.

The Honest Trade-Offs

The cloud is not free lunch. You are placing trust — and it is worth naming in what:

  • Connectivity — no internet, no files (offline modes help; not everything has one)
  • The provider — their outage is your outage; their security breach is your problem; their price rise or shutdown is your migration
  • Privacy and control — your data sits under another organisation's roof and jurisdiction; the safety habits protecting your account (strong password, 2FA) suddenly guard everything
  • Lock-in — years of files and setups make leaving a provider genuinely hard

None of this argues against the cloud — it argues for using it knowingly: keep independent backups of the irreplaceable, secure the account like the master key it is.

Try It Yourself

Audit your own cloud footprint: list five services you use where the data is not on your device. For each — what would you lose if you were locked out tomorrow, and does anything irreplaceable have a second, non-cloud copy? (If the answer alarms you, you know today's homework.)

Worked Example — Why "The Cloud" Handled a Ticket Sale Nobody Owns Servers For

A useful way to see elasticity concretely: a mid-sized band announces a reunion tour, and their ticket website — normally handling a few hundred visitors a day — suddenly needs to handle 200,000 people hitting "buy" in the same ten-minute window. Trace what elasticity actually does here, step by step:

  1. Under normal traffic, the site runs on a handful of rented servers — cheap, appropriately sized for everyday demand
  2. As the on-sale moment approaches, the cloud provider's system detects rising traffic and automatically provisions (rents, on the site's behalf) dozens or hundreds of additional servers within minutes
  3. The surge is absorbed across all those servers simultaneously — this is exactly why a database sits underneath, coordinating which seat sells to whom without two people buying the same seat
  4. Within an hour, demand drops back to normal, and the extra servers are automatically released — the site stops paying for capacity it no longer needs

Owning enough physical servers to survive that ten-minute spike, then leaving them idle for the following 364 days, would be financially absurd for a mid-sized ticket seller — precisely the "guess high, guess low" dilemma described earlier in this lesson. Elastic, rented cloud capacity is what makes a temporary, extreme spike in demand survivable without a permanent, wasteful investment in hardware.

Key Words

  • Cloud computing — computing resources delivered over the internet, rented not owned
  • Data centre — the building full of servers where "the cloud" physically lives
  • Elasticity / scaling — expanding and shrinking rented capacity on demand
  • Replication / redundancy — multiple copies so single failures lose nothing
  • Lock-in — the growing cost of leaving a provider
Related Topics