The server in the closet
Rohan ran a small accounting firm. In 2015, "IT infrastructure" meant one specific thing: a beige server tower humming in a locked closet down the hall, running the firm's client database and email. When it worked, nobody thought about it. When it didn't, the whole firm stopped.
That server cost real money before it did anything useful: buying it, installing it, cooling the closet, paying someone to patch it every month, and buying a second one "just in case" the first one died. Rohan's firm had sixty employees. The server was sized for two hundred, because guessing wrong in the other direction meant the database grinding to a halt during tax season. Most of that capacity sat unused, paid for anyway.
This is the situation cloud computing was built to fix.
Renting instead of owning
Cloud computing means using computing power, storage, and software over the internet, provided by someone else, instead of buying and running the physical hardware yourself. Azure, AWS, and Google Cloud are the three big providers doing the owning; everyone else rents from them.
The clearest way to picture this: it's the difference between owning a car and taking a taxi. Own a car, and you've paid for it whether you drive it or not, insurance, maintenance, a parking spot, all fixed costs. Take a taxi, and you pay only for the ride you actually take, and if you suddenly need five taxis at once for a group trip, that's just as available as needing one. Rohan's server was the car sitting in the driveway sized for a trip he'd take twice a year.
Whose job is it, exactly
Moving to the cloud doesn't mean Microsoft takes care of everything. It's a split, called the shared responsibility model. Microsoft is responsible for the physical stuff you'll never see: the data center building, the power, the physical servers, the network hardware, keeping all of it running and secure. You're still responsible for your side: your data, who has access to it, and depending on what you're using, how your applications are configured.
Where exactly that line sits shifts depending on what kind of service you're using, which is exactly what the next module (service models: IaaS, PaaS, SaaS) is about. But the core idea holds everywhere: the cloud provider secures the cloud itself, you secure what you put in it.
Public, private, and hybrid: where the servers actually sit
"The cloud" isn't one single thing everyone shares identically. There are three models:
- Public cloud — resources live in the provider's data center, shared infrastructure, accessed over the internet. This is what most people mean by "the cloud," and it's Azure's default mode.
- Private cloud — cloud-style resources dedicated entirely to one organization, sometimes in their own data center, for cases with strict regulatory or security requirements.
- Hybrid cloud — a mix of both, some resources on-premises, some in the public cloud, connected together. A common real-world pattern for companies mid-migration, or ones with a specific system that legally has to stay on their own hardware.
Rohan's firm, if it moved today, would almost certainly land on public cloud: no regulatory reason to keep anything on-site, and the cost savings of shared infrastructure are exactly what a sixty-person firm needs.
Pay for what you use, not what you might need
This is the other half of the taxi-versus-car idea, formalized: the consumption-based model. Instead of buying a server sized for your busiest possible day, you pay for compute, storage, and services as you actually use them, and that usage can scale up automatically when demand spikes and back down when it doesn't. Tax season traffic spike, handled. February's quiet stretch, no cost for capacity nobody's using.
This single shift, from a large upfront purchase to ongoing, usage-based spend, is the financial reason cloud adoption happened as fast as it did, well before "the cloud" became a buzzword.
Check your understanding
Five questions, all from this module. Get 4 out of 5 to mark it complete.
1. What was the core limitation of Rohan's on-premises server?
2. In the shared responsibility model, who is always responsible for the physical data center and hardware?
3. A company keeps one regulated system on its own hardware while running everything else in Azure. What model is this?
4. What does "consumption-based pricing" mean?
5. Which is the best analogy for public cloud versus owning your own server, as used in this module?