This domain is 24% of the scored exam. Questions are short and conceptual: “which benefit does this describe?”, “which pillar covers this?”, “which migration strategy is this?”. There are no calculations. You need the vocabulary and the ability to match a one-line scenario to a term.
Benefits of the AWS Cloud
The cloud means renting IT resources over the internet and paying for what you use, instead of buying and running your own data centre.
| Benefit | What it means | Typical wording |
|---|---|---|
| Trade fixed expense for variable expense | No up-front hardware purchase. You pay per hour, second, GB or request | “Avoid large capital investment”, “pay only for what you use” |
| Economies of scale | AWS buys at huge volume, so its unit cost is lower and prices to customers fall over time | “Lower pay-as-you-go prices”, “AWS has reduced prices many times” |
| Stop guessing capacity | Scale up or down to match real demand, instead of over-buying for a peak or running out | “Unpredictable traffic”, “no idle servers” |
| Increase speed and agility | Resources are ready in minutes, so experiments are cheap and teams ship faster | “Launch in minutes”, “try new ideas quickly” |
| Stop spending money running and maintaining data centres | Focus on your own applications, not racks, power and cooling | “Focus on differentiating work” |
| Go global in minutes | Deploy in many Regions close to customers | “Lower latency for users worldwide” |
Elasticity, scalability, agility, high availability
These four are easy to mix up.
| Term | Meaning | Example |
|---|---|---|
| Elasticity | Resources grow and shrink automatically with demand | Auto Scaling adds instances at lunchtime and removes them at night |
| Scalability | The system can handle more load by adding resources (up or out) | A database moved to a bigger instance |
| Agility | Speed of provisioning and changing | New environment in minutes |
| High availability | The workload keeps running when a component fails | Instances in two Availability Zones behind a load balancer |
| Fault tolerance | Continues with no interruption when something fails (a stricter idea than high availability) | Redundant components with automatic failover |
Scaling up (vertical) means a bigger machine. Scaling out (horizontal) means more machines. Elasticity is the automatic, two-way version.
Global infrastructure as a benefit
A Region is a geographic area with several isolated Availability Zones (AZs). Deploying across AZs gives high availability. Deploying in several Regions gives lower latency for distant users, disaster recovery and data residency. The detail of Regions, AZs and edge locations is in the technology domain notes.
Well-Architected Framework
The AWS Well-Architected Framework is a set of best practices for designing and reviewing workloads. It has six pillars. The AWS Well-Architected Tool (a free service in the console) lets you review a workload against the pillars and gives an improvement plan.
| Pillar | Core idea | Cues |
|---|---|---|
| Operational excellence | Run and monitor systems and keep improving the processes | Runbooks, automate changes, small reversible changes, learn from failures, infrastructure as code |
| Security | Protect data, systems and assets | Least privilege, strong identity, traceability, encrypt in transit and at rest, automate security checks |
| Reliability | Recover from failure and meet demand | Test recovery, scale horizontally, automatic recovery, manage change |
| Performance efficiency | Use resources efficiently as demand and technology change | Right resource types, serverless, experiment more often, go global in minutes |
| Cost optimization | Deliver value at the lowest price | Pay for what you use, measure efficiency, stop undifferentiated heavy lifting, analyse spend |
| Sustainability | Reduce the environmental impact of running workloads | Maximise utilisation, use efficient hardware and managed services, reduce downstream impact |
Sustainability was added in 2021. If a question lists six pillars, it includes it. Older material shows only five.
Design principles you may see in a stem:
- Stop guessing capacity, test at production scale, automate to make experimentation easy, allow evolutionary architectures, drive architecture with data, and improve through game days.
- Loose coupling: components depend on each other as little as possible, for example by putting a queue (SQS) between them. A failure in one part doesn’t take down the others.
- Think parallel: spread work across many small resources rather than one large one.
- Disposable resources: treat servers as replaceable, built from automation, not hand-configured pets.
- Design for failure: assume anything can break and plan recovery.
Trap: “Which pillar?” questions often describe a mix. Pick the one the main goal names. Encrypting data is security, restoring after failure is reliability, shutting down idle resources is cost optimization.
Migration to the cloud
AWS Cloud Adoption Framework (AWS CAF)
The AWS CAF is guidance for organisations planning a move to the cloud. It organises the work into perspectives, each owned by a group of stakeholders. There are six: business, people, governance, platform, security and operations.
| Perspective | Concerned with |
|---|---|
| Business | Making sure cloud investment links to business outcomes; strategy, portfolio, innovation, analytics |
| People | Culture, training, organisational change, roles and skills (HR and staffing) |
| Governance | Managing cloud initiatives, aligning them with strategy, programme and risk management, cost control |
| Platform | Architecture and building the technology (infrastructure, data, applications) |
| Security | Confidentiality, integrity and availability; identity, detection, incident response |
| Operations | Day-to-day running, monitoring, change and recovery of workloads |
Benefits the exam names for adopting the CAF: reduced business risk, better environmental, social and governance (ESG) performance, increased revenue and higher operational efficiency. The CAF is guidance, not a tool you run.
The migration strategies (“the 7 Rs”)
| Strategy | Meaning | Example |
|---|---|---|
| Retire | Switch off what’s no longer needed | Unused application |
| Retain | Keep it where it is for now | Legacy system not ready |
| Rehost (“lift and shift”) | Move as is to EC2, with no changes | Quick move of a VM |
| Relocate | Move to the cloud version of the same platform without changes to hypervisor-level setup | VMware workloads to VMware Cloud on AWS |
| Repurchase (“drop and shop”) | Replace with a different product, often SaaS | Own CRM to Salesforce |
| Replatform (“lift, tinker and shift”) | Small optimisations, no core change | Self-run database to Amazon RDS |
| Refactor / re-architect | Redesign to use cloud-native features | Monolith to microservices on Lambda |
Rehost is the fastest and least change. Refactor takes the most effort and gains the most cloud benefit.
Migration services and data transfer
| Need | Service |
|---|---|
| Discover on-premises servers, dependencies and utilisation | AWS Application Discovery Service |
| One place to track a migration across tools | AWS Migration Hub |
| Lift and shift servers to EC2 | AWS Application Migration Service |
| Move a database with minimal downtime, replicating changes continuously | AWS Database Migration Service (DMS) |
| Convert a schema between database engines (for example Oracle to PostgreSQL) | AWS Schema Conversion Tool (SCT) or DMS Schema Conversion |
| Move large amounts of data offline in a physical device | AWS Snow Family (Snowcone, Snowball Edge, Snowmobile for exabyte scale) |
| Move data online over the network | AWS DataSync, and AWS Transfer Family for SFTP/FTPS/FTP into S3 or EFS |
| Disaster recovery of servers | AWS Elastic Disaster Recovery |
Rule of thumb: terabytes to petabytes with poor or slow bandwidth means Snow Family. A database replication need means DMS.
Cloud economics
Fixed vs variable cost
- Fixed (capital) expense (CapEx): pay up front for servers, network gear, buildings. You pay whether you use the capacity or not, and you must forecast demand.
- Variable (operating) expense (OpEx): pay as you consume, with no up-front commitment.
On-premises costs people forget
Hardware and its refresh cycle, data-centre space, power and cooling, physical security, networking, staff to operate it, software licences, and the cost of idle capacity bought for peak. The cloud’s total cost of ownership (TCO) comparison counts these. The AWS Pricing Calculator estimates AWS cost for a planned workload. The older “TCO calculator” idea is the same intent.
Licensing: BYOL vs included
| Model | Meaning |
|---|---|
| License included | The software licence (for example Windows Server or SQL Server) is part of the hourly price |
| Bring your own licence (BYOL) | You use licences you already own, so you pay only for the infrastructure |
AWS License Manager tracks licence usage and enforces rules across AWS and on-premises. Dedicated Hosts are often needed for BYOL where the licence is tied to physical sockets or cores.
Rightsizing
Rightsizing is matching instance type and size to the actual need, so you don’t pay for unused capacity. AWS Compute Optimizer analyses utilisation and recommends sizes. Cost Explorer gives rightsizing recommendations too. Rightsizing is a continuous activity, not a one-off.
Automation
Automating with AWS CloudFormation (infrastructure as code) makes environments repeatable, reduces human error, speeds up provisioning and lets you delete a whole stack when finished so you stop paying. It also supports consistent configuration across accounts.
Managed services
A managed service means AWS runs the undifferentiated work: patching, backups, scaling, high availability. Examples named in the guide: Amazon RDS, Amazon ECS, Amazon EKS and Amazon DynamoDB. The benefit is lower operational overhead and often lower TCO, at the cost of less control of the underlying machine.
Cost-saving levers in one place
| Lever | How it saves |
|---|---|
| Pay as you go | No wasted capacity |
| Volume discounts | Tiered pricing gets cheaper per unit as usage rises (for example S3 and data transfer tiers) |
| Savings Plans, Reserved Instances | Commit for one or three years for a discount |
| Spot Instances | Spare capacity at a big discount, can be interrupted |
| Rightsizing and Auto Scaling | Remove idle capacity |
| Managed services and serverless | Less operational effort |
Exam checklist
- Six advantages of cloud computing: variable instead of fixed expense, economies of scale, stop guessing capacity, speed and agility, no data-centre upkeep, go global in minutes.
- Elasticity = automatic scale out and in. High availability = survive component failure (multiple AZs).
- Six Well-Architected pillars: operational excellence, security, reliability, performance efficiency, cost optimization, sustainability. The Well-Architected Tool reviews workloads.
- CAF has six perspectives: business, people, governance, platform, security, operations.
- Rehost = lift and shift to EC2. Replatform = small tweak (database to RDS). Refactor = rebuild cloud-native. Repurchase = switch to SaaS. Retire, retain, relocate complete the list.
- Snow Family for offline bulk transfer; DMS for databases; SCT for schema conversion; Migration Hub to track.
- BYOL vs license included; Dedicated Hosts for socket-bound licences; License Manager to track.
- Rightsizing is done with Compute Optimizer and Cost Explorer.
- CloudFormation is the example of IaC and automation. RDS, ECS, EKS and DynamoDB are examples of managed services.