sdlc.dev

AWS fundamentals path · Ground rules · Topic 1 of 18

Regions and Availability Zones

Not started

AWS is not one big computer. It is a set of **Regions** (such as eu-west-1 in Ireland), each a separate group of data centres, and inside each Region a few **Availability Zones** (AZs). A zone is one or more data centres with independent power, cooling and network links, placed far enough from the other zones in the Region that a fire, flood or power failure should hit only one of them. Almost every design decision about availability on AWS comes down to which of these boundaries a resource sits inside.

Why it matters: Every AWS resource lives in a Region, and most live in a single zone inside it. Knowing which is which tells you what a data centre failure will take down.

  1. 1A Region is a separate world

    Each Region is an independent set of data centres with its own copy of most services. Resources, quotas and prices are per Region.

  2. 2A zone is a failure boundary

    An Availability Zone is one or more data centres with their own power and networking, far enough from the others to fail alone.

  3. 3Know each service's scope

    An EC2 instance or EBS volume is zonal, S3 and DynamoDB are regional, IAM is global. The scope decides what one zone failing breaks.

  4. 4Spread across zones by default

    Running in at least two zones (three is better) is the cheapest big step up in availability AWS offers.

1

How it works

The explanation, with examples you can run.

Three kinds of scope

ScopeExamplesWhat a single failure takes down
ZonalEC2 instances, EBS volumes, a subnet, a zonal NAT gatewayEverything in that zone
RegionalS3, DynamoDB, SQS, Lambda, an ALB spread over several zonesRarely anything; the service is built across zones
GlobalIAM, Route 53, CloudFrontDesigned with a separate control plane and data plane

A regional service such as S3 already spreads your data across zones for you. A zonal resource does not: if you run one instance, you run it in one zone, and it shares that zone's fate. That is why the default advice is to run every tier in at least two zones, and three where the service supports it, so that losing a zone means losing a third of your capacity rather than all of it.

Data planes and control planes

AWS separates the control plane (the APIs that create, change and delete resources) from the data plane (the resources doing their job: an instance serving requests, a queue delivering messages). Data planes are simpler and are designed to keep working when control planes struggle. The practical lesson, which the AWS fault-isolation whitepaper calls static stability, is to avoid recovery plans that depend on calling a control plane during an outage. If you need three zones' worth of capacity to survive one zone failing, provision it in advance rather than counting on launching new instances in the middle of the event, when everyone else is trying to do the same.

Choosing a Region

Pick a Region for the latency to your users, the data residency rules you are under, which services and instance types are offered there (newer services arrive in some Regions first), and price, which differs between Regions. Quotas are per Region too, so a second Region needs its own limit increases. Most Regions have three or more zones; a few older ones offer fewer to new accounts.

Zone names are not always zone IDs

In AWS's oldest Regions, accounts created before November 2025 have their own mapping of zone names to physical zones, so us-east-1a in one account may be a different building from us-east-1a in another. Newer accounts get the same mapping. The zone ID (use1-az1) is the same everywhere. Use zone IDs when two accounts need to agree on placement, for example to keep cross-zone traffic (which is billed) to a minimum.

2

Check yourself

3 questions. Get them all right to finish this topic.

  1. 1.You run one EC2 instance with an EBS volume in eu-west-1a. The zone has a power event. What survives?
  2. 2.Why can services replicate synchronously between zones but usually not between Regions?
  3. 3.Two accounts both run in us-east-1a. Are they guaranteed to be in the same physical zone?

0 of 3 right so far. Get all three to finish this topic.

Go deeper