What Is Cloud Computing? IaaS, PaaS, SaaS and Deployment Models Explained

Think about the IPL final night. A streaming app goes from a few lakh viewers to several crore in an hour, then drops back after the last ball. If that company had to buy enough servers for the final, most of them would sit idle for the rest of the year. Instead, it rents capacity from a cloud provider for exactly the hours it needs — and returns it when the match ends.

That is cloud computing: on-demand access to computing resources — servers, storage, databases, networking, software — over the internet, paying only for what you use.

Before the cloud

A startup in 2008 that wanted to launch an app had to buy servers, rent space in a data centre, arrange power, cooling and internet links, install everything, and guess how much capacity it would need a year later. Guess too low and the site crashed on launch day; guess too high and money sat in idle hardware. Setting up could take weeks.

Today the same startup can have servers running worldwide in minutes, with a credit card.

Five characteristics that make something "cloud"

1. On-demand self-service — create a server yourself in minutes, no ticket to an IT team.
2. Broad network access — use it from anywhere over the internet.
3. Resource pooling — the provider shares huge data centres across many customers.
4. Rapid elasticity — scale up for the IPL final, scale down afterwards, often automatically.
5. Measured service — pay per hour, per gigabyte, per request, like an electricity meter.

These come from the NIST definition of cloud computing, which interviewers sometimes quote.

The three service models

The easiest way to understand them is the "pizza" comparison: make it at home, buy a ready base, order delivery or eat at a restaurant — the more you pay someone else to handle, the less you manage.

You have been a SaaS user for years — every time you open Gmail. Developers mostly work with IaaS and PaaS, and increasingly with serverless (functions that run only when called), which we cover later.

Deployment models

• Public cloud — shared infrastructure from AWS, Microsoft Azure, Google Cloud, Oracle Cloud and others. Most startups and many enterprises.
• Private cloud — cloud-style infrastructure used by one organisation only, on its own premises or hosted. Common where regulation or control matters (some banks, government).
• Hybrid cloud — a mix: sensitive systems stay private, everything else runs in public cloud, connected securely.
• Multi-cloud — using more than one public provider, to avoid lock-in or use each one's strengths.

Why companies move to the cloud — and the catch

Benefits: no upfront hardware cost (capital expense becomes operating expense), scaling in minutes, global reach, managed services that save engineering time, and strong built-in reliability options.

The catch: costs can grow quietly if nobody watches them, misconfiguration is a leading cause of data leaks, and deep use of one provider's special services makes switching harder (vendor lock-in). The last lesson of this track is about controlling exactly these risks.

The big providers

AWS (Amazon Web Services) has the largest market share and the widest range of services. Microsoft Azure is strong with enterprises already using Microsoft software. Google Cloud is popular for data, analytics and AI workloads. All three have data centre regions in India (Mumbai, Hyderabad, Pune, Delhi and others), which matters for speed and data-residency rules.