---
title: "Cloud foundations and global infrastructure"
chapter: "01"
---

# Cloud foundations and global infrastructure

AWS rents computing building blocks over a network. You combine those
blocks into a workload and pay for what you use.

## The map

- A **Region** is a separate geographic area.
- An **Availability Zone (AZ)** is an isolated location inside a Region, made
  from one or more data centers with separate power and networking.
- An **edge location** puts content and network services close to users.
- An **AWS account** is a strong boundary for billing, permissions, quotas, and
  blast radius.
- A **resource** is something you create: a bucket, instance, database, queue,
  function, key, or network.

## The simple design rule

Use multiple AZs for high availability inside one Region. Use multiple Regions
only when business needs justify the extra complexity: strict disaster recovery,
data residency, or globally low latency.

## Shared responsibility

AWS secures **the cloud**: physical facilities, hardware, and the managed service
platform. The customer secures what is placed **in the cloud**: identities, data,
configuration, application code, and many operating-system duties. The exact
line changes by service. With EC2 you manage more. With Lambda or DynamoDB AWS
manages more of the underlying stack.

## Managed services first

Choose the highest-level managed service that meets the requirement. It usually
removes undifferentiated work such as patching, replication, and replacement.
Do not choose it blindly: check service limits, portability, skills, latency,
features, and cost at the expected scale.

## Architect questions

1. Where are users and regulated data?
2. Which failures must the design survive?
3. What is the tolerated latency?
4. What must remain on premises?
5. Which teams operate the solution?
6. How will accounts and environments be separated?

## Common trap

“The cloud is automatically highly available” is false. AWS supplies resilient
building blocks. You must place and configure them so the workload survives the
failures that matter.