What Is AWS?

Cloud computing before the product names

AWS is far easier to learn when the delivery model comes first and the product names come second. Almost nothing in this certification is a question about a machine; the questions are about how capability is obtained, paid for, and placed.

Cloud computing is the on-demand delivery of IT resources over the Internet with pay-as-you-go pricing. Two promises hide inside that sentence: the resource appears when it is requested, and the bill follows what was actually used.

Instead of buying, owning, and maintaining physical data centers and servers, customers can access compute, storage, and database services as needed. Computing power, storage, and databases become services a team requests, not equipment a team owns.

Notice what does not change. The organization still decides which resources its application needs and how they are configured. What disappears is the procurement project in front of that decision: traditionally, everything from server hardware and storage systems to networking and security technology had to be bought and configured before a single digital system could launch.

Organizations of every type, size, and industry work this way — for data backup, disaster recovery, email, virtual desktops, software development and testing, big data analytics, and customer-facing web applications.

Capacity that follows demand

Elasticity is the first reason this model changes architecture rather than just accounting.

AWS cloud services can scale capacity up or down quickly while customers pay only for what they use, and complete control of those computing resources stays with the customer. Cloud capacity can be adjusted as demand changes instead of being fixed by an initial hardware purchase.

A purchase forces the sizing question to be answered once, in advance — and it is easy to be wrong in both directions. Too small, and the application fails at its peak. Too large, and the surplus is paid for all year.

Variable expenses are also lower than doing it yourself, because a provider reaches economies of scale across many customers.

Elasticity answers how much. It says nothing about where, which is a separate decision later in this concept.

What AWS provides inside that model

With the model clear, AWS is simply the provider that supplies capability inside it.

AWS provides cloud and AI capabilities on a global infrastructure designed for security, reliability, and performance. That infrastructure is built for scalability and performance, and security is treated as the first requirement rather than a later addition.

AWS infrastructure supports applications running in the cloud, on premises, or at the edge. A workload therefore does not have to live entirely in one place to use the platform: a design can keep part of the system in a customer facility, or close to where its data is produced.

Capabilities are combined, not bought whole

The most common beginner mistake is to read AWS as one product with one configuration.

AWS capabilities can be combined to build solutions that balance performance and cost-effectiveness. The breadth exists precisely so that a design can be assembled from parts that each answer a requirement.

The working order for an architect is therefore:

1. State the requirements and constraints of the workload.
2. Choose the capabilities that satisfy them.
3. Combine those capabilities into a single solution.
4. Re-check the result against every stated constraint, cost included.

That sequence is why two designs that both "work" can still differ in quality. Both may use real services in supported ways; the better one satisfies more of the stated constraints at the same time.

Combining capabilities does not remove the need to understand the workload. It rewards understanding it.

How much of the stack you control

Cloud does not mean that every service exposes the same operating boundary. The service model tells you how much of the stack the customer still runs.

The three main cloud computing types are Infrastructure as a Service, Platform as a Service, and Software as a Service. Each cloud computing type provides a different level of control, flexibility, and management, so a team can select the right set of services for its needs.

This is why two feasible designs can still be different answers. One keeps control and accepts the management work; the other sheds the management work and gives up control. Neither is universally right, so the deciding factor is the level of control and operating responsibility the situation genuinely requires.

The building blocks of Infrastructure as a Service

Infrastructure as a Service provides basic building blocks such as networking, computers, and data storage. The computers may be virtual or on dedicated hardware, and the storage is space the customer organizes for itself.

Underneath those blocks, AWS manages the underlying infrastructure that makes cloud services available on demand. All the required hardware, operating systems, and other infrastructure layers are held and maintained in secure AWS data centers around the world — which is why cloud services are also called cloud managed services.

So IaaS is the model to reach for when a team must choose its own compute and network components directly, and it still requires no data center of its own. Combine that with elastic capacity and a single design can serve unpredictable peaks while keeping direct control of its building blocks.

Where the workload meets its users

Capacity answers how much. Placement answers where, and it is a separate design dimension with its own symptoms.

With the cloud, a team can expand into new geographic regions and deploy globally in minutes, because the provider already operates infrastructure all over the world. Deploying applications closer to end users can reduce latency and improve their experience.

Read every scenario with both dimensions in mind. A design that scales beautifully in one distant location has answered only half of a problem that mentions faraway audiences and sharp demand swings.

Key takeaways

• Cloud computing delivers IT resources on demand over the Internet with pay-as-you-go pricing, so capability is accessed instead of physical data centers and servers being bought and maintained.
• AWS provides cloud and AI capabilities on a global infrastructure and expects them to be combined into solutions that balance performance and cost-effectiveness, in the cloud, on premises, or at the edge.
• IaaS, PaaS, and SaaS provide different levels of control, flexibility, and management; IaaS exposes the basic blocks — networking, computers, and data storage — while AWS manages the infrastructure beneath them.
• Elastic capacity and global placement solve different constraints: capacity scales with demand, and placing an application closer to its users reduces latency.