Cloud Computing and Service Models (IaaS/PaaS/SaaS)

Build the mental model

Cloud computing means using computing resources — servers, storage, databases, networking — provided over a network by someone else, rather than owned on your own hardware. "Someone else's computer" is a common joke, and a reasonable starting point.

But that undersells it. Cloud platforms provide managed services — databases that handle their own backups, storage built to survive hardware failure, networking a specialized team would otherwise need to build. You're renting expertise, not just a machine.

Model	Control vs Convenience
IaaS	Most control — you manage the OS and runtime; typical use: custom server setups, specialized workloads.
PaaS	Balanced — deploy code, the platform manages servers and scaling; typical use: most web applications.
SaaS	Most convenience — use a finished app, nothing to manage; typical use: email, CRM, support tools.

Moving from IaaS toward SaaS trades control for convenience, and neither end is universally correct — the right choice depends on what a project actually needs to manage itself.

IaaS: Infrastructure as a Service — rents raw computing building blocks (virtual servers, storage, networking); you manage the OS and everything above it.

PaaS: Platform as a Service — you deploy application code; the platform manages the underlying servers, OS, and scaling automatically.

SaaS: Software as a Service — you use a finished application over the internet, with no infrastructure of any kind to manage.

text

IAAS TO SAAS: ABSTRACTION INCREASES
-----------------------------------
[MORE CONTROL, MORE WORK]                [LESS CONTROL, LESS WORK]

    IaaS      -------->      PaaS      -------->      SaaS
(servers,               (deploy code,            (use the app,
 storage,                 platform                 nothing to
 networking)              handles rest)            manage)

Connect it to a real scenario

Choosing a model usually comes down to operational responsibility. Deep control over the OS or specific versions points to IaaS; writing code without configuring servers points to PaaS; needing a capability rather than a platform points to SaaS.

Most real organizations mix models rather than picking one for everything — core application on PaaS, several SaaS tools for internal operations, one specialized workload on IaaS because it needs a configuration no managed platform offers.

Do you need control over the OS or specific software versions? -> IaaS
Do you just want to deploy code and let the platform handle servers? -> PaaS
Do you just need a working tool, not a platform to build on? -> SaaS

Cloud computing means renting managed infrastructure and expertise, not just a remote machine.
IaaS, PaaS, and SaaS trade control for convenience in increasing steps.
Most real organizations mix service models rather than committing to just one.

Try the working example

javascript

function recommendModel(wantsToManage) {
  if (wantsToManage === "servers-and-os") return "IaaS";
  if (wantsToManage === "just-my-code") return "PaaS";
  return "SaaS";
}

const preferences = ["servers-and-os", "just-my-code", "nothing-just-use-the-app"];

for (const pref of preferences) {
  console.log(`${pref} -> ${recommendModel(pref)}`);
}

You should see

servers-and-os -> IaaS
just-my-code -> PaaS
nothing-just-use-the-app -> SaaS

5-minute try-it

Add a fourth preference string, such as 'i-just-need-email', to the preferences array and extend recommendModel() with one more condition so it correctly returns 'SaaS' for it without breaking the existing three cases.

One important caution

Choosing IaaS for a simple web app that would run fine on PaaS, taking on server maintenance work that buys nothing.

Assuming SaaS tools require zero decisions — data ownership, export options, and vendor lock-in still matter.

NIST SP 800-145: The NIST Definition of Cloud Computing — Cloud & Deployment