Every section below carries a Source line. The prose is written for this guide; the facts come from the page cited. Where a fact could not be verified against official documentation it is marked [UNVERIFIED] rather than guessed — those are the lines to check before this ships.
AWS frames the case for cloud computing as six advantages, and the exam draws on them repeatedly. They are worth knowing as a set, because several exam questions are really asking "which of these six is this scenario describing?"
Trade fixed expense for variable expense. Running your own data centre means buying servers before you know how much you will use them. In the cloud you pay for what you consume, when you consume it. The money moves from capital expenditure — a large purchase up front — to operating expenditure, a running cost that tracks actual usage.
Benefit from massive economies of scale. Because a cloud provider aggregates the usage of hundreds of thousands of customers, it buys hardware, power and bandwidth at a scale no single organisation can match. Those savings reach the customer as lower pay-as-you-go prices.
Stop guessing capacity. Deciding how much infrastructure to buy before an application launches leaves you either paying for idle machines or running out of room. Cloud capacity can be increased or decreased within minutes, so the guess is no longer necessary.
Increase speed and agility. New resources are available in minutes rather than the weeks a hardware purchase takes. The consequence is not only speed but willingness to experiment: when trying something costs little and takes minutes, an organisation attempts more things.
Stop spending money running and maintaining data centres. Racking, stacking and powering servers is work that does not distinguish your business from any other. Moving it to a provider frees the effort for what does.
Go global in minutes. An application can be deployed in several Regions around the world without building anything in those places, which puts it closer to users and lowers latency for them.
Three words the exam uses precisely, and confuses candidates who treat them as synonyms.
Elasticity is capacity following demand. Resources are added when load rises and removed when it falls, so you are paying for roughly what you are using rather than for a peak that occurs twice a year.
High availability is the system staying usable when part of it fails. It is achieved by running across more than one isolated location, so the loss of one does not take the service down.
Agility is organisational rather than technical: the speed at which you can try, measure and discard ideas. It is a consequence of the other two plus the absence of procurement.
The distinction to hold on to: elasticity is about cost tracking demand, availability is about surviving failure, agility is about speed of change.
The Well-Architected Framework is AWS's set of design principles for building systems on its platform. It is organised into six pillars, and the exam expects you to name them and tell them apart.
| Pillar | The question it answers |
|---|---|
| Operational excellence | Can we run and monitor this, and keep improving how we do it? |
| Security | Are the data and the systems protected? |
| Reliability | Does it recover from failure and keep meeting demand? |
| Performance efficiency | Are we using the right resources, and still the right ones as things change? |
| Cost optimization | Are we getting this outcome at the lowest price? |
| Sustainability | What is the environmental cost of running it? |
Two things candidates get wrong. Sustainability is the newest pillar and is frequently missing from older study material — if a question offers five pillars, look for it. And performance efficiency is not the same as reliability: a system can be fast and fragile, or slow and dependable.
AWS also provides the AWS Well-Architected Tool, a service that reviews a workload against these pillars and reports where it falls short. The Framework is the ideas; the Tool is the thing you run.
Moving to the cloud is not only a technical project, and AWS CAF exists to say so. It organises the work into six perspectives, each owned by different people in an organisation. The first three are about the business; the last three are about the technology.
| Perspective | Concern |
|---|---|
| Business | Making sure cloud investment actually advances business outcomes |
| People | Culture, structure, leadership and skills — the bridge between technology and business |
| Governance | Orchestrating initiatives so benefits are maximised and transformation risk minimised |
| Platform | Building the scalable hybrid platform, modernising existing workloads, building new ones |
| Security | Confidentiality, integrity and availability of data and workloads |
| Operations | Delivering cloud services at the level the business needs |
The exam tends to describe a problem and ask which perspective owns it. A training gap is People. An unclear approval process for spending is Governance. Encryption of stored data is Security.
There are seven strategies for moving an application to the cloud, known as the 7 Rs. The exam describes a situation and asks which one fits, so the useful thing is not the list but the distinctions between neighbours.
| Strategy | What it means |
|---|---|
| Rehost (lift and shift) | Move the application to the cloud without changing it to take advantage of cloud capabilities |
| Relocate (hypervisor-level lift and shift) | Move infrastructure to the cloud without buying new hardware, rewriting applications or changing existing operations |
| Replatform (lift and reshape) | Move it, and introduce some optimisation along the way to use cloud capabilities |
| Repurchase (drop and shop) | Switch to a different product, typically replacing a traditional licence with a SaaS subscription |
| Refactor / re-architect | Move it and change its architecture, using cloud-native features for agility, performance and scalability |
| Retain (revisit) | Leave it where it is for now — major refactoring you want to postpone, or legacy systems with no business case to move |
| Retire | Decommission it; it is no longer needed |
The three pairs that get confused:
Rehost against Relocate. Both move things unchanged. Rehost moves the application; relocate moves infrastructure at the hypervisor level, so you are not buying hardware, rewriting anything, or altering how you operate.
Replatform against Refactor. Both change something. Replatform makes a limited optimisation — moving a self-managed database to a managed one. Refactor rebuilds the architecture around cloud-native features. Refactor delivers the most, costs the most in skills and time.
Retain against Retire. Retain keeps it running where it is; retire switches it off. A question mentioning an application nobody uses any more is retire; one mentioning a system too complex to move yet is retain.
The trade-off AWS states plainly: rehosting is fastest and least disruptive but gains the least, and carries a pre-existing platform problem across unchanged. Refactoring is the most cost-effective in the end but needs advanced skills and takes longest.
The pattern the exam rewards: Discovery finds what you have, Evaluator prices it, Hub tracks the work, and the Migration/Database services do it.
A fixed cost is paid whether or not anything is used — a purchased server depreciates in a rack at three in the morning exactly as it does at midday. A variable cost moves with consumption. The central economic claim of cloud computing is that it converts the first into the second, which is what makes "stop guessing capacity" financially meaningful rather than merely convenient.
Candidates under-count this, and so do exam distractors. Beyond the price of the hardware there is the building, power, cooling, physical security, network connectivity, the staff who maintain it, the spare capacity held for growth, and the replacement cycle when the hardware ages out.
Some software can be brought to AWS under a licence you already own, which is the Bring Your Own License (BYOL) model. Alternatively, AWS offers instances where the licence cost is included in the hourly price. BYOL suits an organisation with existing agreements it has already paid for; included licensing suits one that would otherwise have to buy them. AWS License Manager exists to track licence usage against the terms you hold.
Rightsizing is matching the resource to the actual workload — moving from an instance that is twice the size the work needs to one that fits. It is the most common single source of cloud savings, because capacity chosen at launch is almost never the capacity the workload settles into. AWS Compute Optimizer analyses usage and recommends the change.
Automating provisioning and configuration reduces cost in two ways that are easy to conflate: less human time spent on repetitive work, and fewer mistakes that cost money — the forgotten test instance running for a month, the environment built slightly differently from the last one.
The same idea as the cloud advantage above, seen from the customer's side: you pay prices that only enormous aggregate purchasing can produce, without having to be enormous.
Practise this with AWS Cloud Practitioner questions