## glossary

What is Cloud Computing?

Cloud computing is the delivery of technology resources — servers, storage, databases, networking — on demand over the internet, billed in proportion to usage. Instead of buying machines and maintaining them, you rent capacity and hand it back when you no longer need it.

The shift is economic before it is technical: capital expenditure becomes operational expenditure. There is no need to forecast three years of demand and buy hardware for the peak — you adjust in minutes.

The three service models

IaaS (Infrastructure as a Service) — you rent virtual machines, disk and network, and manage the operating system and the application.
PaaS (Platform as a Service) — you hand over the code and the platform takes care of servers, scaling and updates.
SaaS (Software as a Service) — you consume the finished software, managing nothing.

The rule of thumb is simple: the higher up the stack, the less control and the less operational work. Small teams win by going to PaaS; teams with specific compliance or performance requirements go down to IaaS.

Deployment models

The public cloud shares infrastructure among customers and offers the best unit cost. The private cloud is dedicated, adopted for regulatory or latency requirements. The hybrid model combines the two, keeping sensitive data in-house and elasticity outside. And multicloud means using more than one provider, usually to reduce dependency or take advantage of specific services.

What changes in the architecture

The cloud rewards applications designed for it. Servers should be treated as cattle, not pets — if one fails, another takes its place. State lives outside the instance, in a database or object storage. And scaling means adding instances, not making the machine bigger.

Elastic scaling declared by metric

# the infrastructure reacts to demand without intervention
minimum: 2
maximum: 20
rule:
  metric: avg_cpu
  target: 65%
  scale_up_when: "> 70% for 3 minutes"
  scale_down_when: "< 30% for 10 minutes"

The risks nobody mentions at the start

Cost is the most common one: the ease of spinning up resources leads to spending that grows unchecked, and the fee for transferring data out of the provider tends to surprise. Hence the discipline of FinOps. There is also lock-in — the more proprietary services you use, the more expensive it gets to leave. Practical comparisons between providers are in comparing cloud services and servers and the AWS universe.

## faq

Frequently asked questions

Is the cloud cheaper than owning servers?

It depends on the usage profile. For variable demand, new projects and small teams, almost always yes. For high, constant, predictable load over years, owned servers can cost less — there are public cases of companies that moved back from the cloud for that reason.

What is serverless?

It is a model where you run functions without managing servers and pay only for execution time, scaling from zero to thousands automatically. Servers still exist — they are just not your responsibility. The typical cost is the latency of the first call, the so-called cold start.

Is my data safe in the cloud?

The major providers offer physical and infrastructure security beyond what most companies can match. The critical point is the shared responsibility model: the provider secures the cloud, you secure what you put in it. Most leaks come from customer misconfiguration.