🎯 By the end of this module you should be able to…
- Define cloud computing and describe the six advantages AWS lists.
- Distinguish IaaS, PaaS and SaaS, and public, private and hybrid deployment.
- Explain the three fundamental drivers of an AWS bill: compute, storage and outbound data transfer.
- Compare On-Demand, Reserved, Savings Plans, Spot and Dedicated pricing.
- Explain Total Cost of Ownership and why on-premises costs are usually understated.
- Describe what AWS Organizations gives you: consolidated billing, SCPs and account separation.
- Navigate the billing tools — Cost Explorer, Budgets, the Pricing Calculator — and pick the right support plan.
What the cloud actually changes
Cloud computing is the on-demand delivery of IT resources over the internet with pay-as-you-go pricing. The technical change is smaller than people think; the economic change is enormous.
Traditional data centre
- Buy hardware months before you need it
- Capacity guessing — you are either wasting money or turning customers away
- Capital expenditure, depreciated over years
- Weeks to provision a server
- You pay for racks, power, cooling, security guards and staff
Cloud
- Provision in minutes, release when finished
- Scale up and back down with demand
- Operating expenditure — a monthly bill
- Someone else runs the building and the hardware
- Go global by choosing another Region
The six advantages of cloud computing
- 💰Trade capital expense for variable expense Pay only for what you consume, when you consume it.
- 📉Benefit from massive economies of scale AWS buys hardware at a scale no single customer can match, and passes some of it on.
- 🎯Stop guessing capacity Scale up and down instead of buying for a peak that may never come.
- ⚡Increase speed and agility Minutes instead of weeks changes what experiments are worth trying.
- 🏢Stop spending money running data centres Spend it on what makes your product different instead.
- 🌍Go global in minutes Deploy to a Region near your users and cut latency without opening an office.
Service and deployment models
| Model | You manage | Provider manages | AWS example |
|---|---|---|---|
| IaaS | OS, runtime, application, data | Virtualisation, servers, storage, network | Amazon EC2, Amazon VPC |
| PaaS | Application and data | Everything below that, including the OS | Elastic Beanstalk, RDS, Lambda |
| SaaS | Your data and users only | The entire stack | Amazon WorkMail, Amazon Chime |
Public cloud is everything in AWS. Private cloud is your own virtualised data centre. Hybrid connects the two — which is what Direct Connect, Storage Gateway and AWS Outposts exist to support.
Introduction to AWS and moving to the cloud
AWS launched in 2006 and now offers well over two hundred services. For this course you only need a mental map of the main categories.
Three ways to talk to AWS
🖱️ Management Console
A browser interface. Best for learning, exploring and one-off tasks. Hard to repeat exactly.
⌨️ AWS CLI
One command per action, scriptable and repeatable. This is what you use in the S3 website lab.
🧱 SDKs & IaC
Python (boto3), Java, JavaScript and so on; plus CloudFormation, CDK and Terraform for describing infrastructure as code.
The migration strategies
| Strategy | Also called | What you do |
|---|---|---|
| Rehost | Lift and shift | Move the server as it is onto EC2. Fastest, least benefit. |
| Replatform | Lift and reshape | Small optimisations — move the database to RDS, for instance. |
| Repurchase | Drop and shop | Replace with a SaaS product. |
| Refactor | Re-architect | Rebuild cloud-native — serverless, managed services. Most benefit, most work. |
| Retire | — | Switch it off. Migrations always find things nobody uses. |
| Retain | Revisit | Leave it where it is, for now. |
Fundamentals of pricing
Compute — what you pay for running instances. Storage — what you pay per GB held. Outbound data transfer — what you pay to send data out of AWS. Data in is generally free, and traffic within an Availability Zone using private addresses is generally free. Almost every surprise bill is one of these three, and usually the third.
| Pricing model | Commitment | Typical saving | Use it for |
|---|---|---|---|
| On-Demand | None — per second or per hour | Baseline | Unpredictable workloads, development, first experiments |
| Reserved Instances | 1 or 3 years, specific instance type | Up to ~72% | Steady, always-on servers |
| Savings Plans | 1 or 3 years, committed $/hour | Up to ~72% | Steady spend, but you want flexibility across instance families |
| Spot Instances | None — AWS can reclaim with 2 minutes' notice | Up to ~90% | Batch jobs, rendering, CI, anything interruptible |
| Dedicated Hosts | Physical server reserved for you | Costs more | Licence terms that require dedicated hardware, strict compliance |
Free Tier — three different kinds
⏳ 12 months free
From the day you open the account. For example 750 hours a month of a t-family EC2 instance and 5 GB of S3 standard storage.
♾️ Always free
Never expires. For example 1 million Lambda requests a month and 25 GB of DynamoDB storage.
🧪 Trials
Short-term, service-specific trials that start when you activate the service.
Exceed it and you are billed at normal rates, silently. Set a zero-spend budget and a billing alarm on the very first day of any student account. This is the single most valuable habit in this module.
Total Cost of Ownership
Comparing a server's price tag with an EC2 hourly rate is not a comparison. TCO counts everything a workload really costs over its life.
| Cost area | On-premises | AWS |
|---|---|---|
| Server hardware | Purchase, plus refresh every 3–5 years | Included in the hourly rate |
| Storage hardware | Purchase, plus spare disks | Per GB per month |
| Network equipment | Switches, routers, firewalls, cabling | Included; you pay for data transfer |
| Facilities | Rack space, power, cooling, physical security | Included |
| Staff | Data-centre technicians, on-call rotations | Largely removed; you still need cloud engineers |
| Software licences | OS, hypervisor, backup, monitoring | Often bundled or metered hourly |
| Idle capacity | Bought for peak, idle most of the year | Scale down and stop paying |
| Disaster recovery | A second site, mostly idle | Another Region, paid only when used |
Utilisation. A typical on-premises server runs at 10–20% utilisation because it was bought for the peak. You paid 100% of it. That single factor is usually the largest line in a TCO comparison — larger than the hardware price itself.
The tool for this
The AWS Pricing Calculator (which replaced the older Simple Monthly Calculator) lets you model a workload service by service and export the estimate as a shareable document. Build the estimate before you build the architecture, not after.
AWS Organizations, billing and support
AWS Organizations
One management account holding many member accounts, grouped into organizational units.
- 🧾Consolidated billing One invoice for everything, and volume discounts pooled across all accounts.
- 🚫Service Control Policies (SCPs) Guardrails that set the maximum permissions an account can ever have. An SCP never grants anything — it only limits. Even the account's root user is bound by it.
- 🏗️Account separation Separate accounts for production, development and sandbox is the cleanest blast-radius control AWS offers.
- 🆕Automated account creation New accounts on demand, with a baseline applied.
The billing tools
📊 Billing Dashboard
Month-to-date spend, a forecast, and a breakdown by service. The first place to look.
🔎 Cost Explorer
Up to 12 months of history, filtered by service, account, Region or tag, with a forecast. This is where you find the one service quietly costing you money.
🔔 AWS Budgets
Set a cost, usage or Savings Plan target and get alerted — or trigger an action — when you approach it.
🏷️ Cost allocation tags
Tag every resource with owner, project and environment, activate the tags, and your bill becomes a report you can hand to a department.
1. Enable MFA on the root user and then stop using it. 2. Create an IAM admin user or role for day-to-day work. 3. Create a zero-spend or small budget with an email alert. 4. Turn on CloudTrail. 5. Pick a Region and stay in it, so you do not leave orphaned resources elsewhere.
Technical support plans
| Plan | Who answers | Response target | Trusted Advisor |
|---|---|---|---|
| Basic (free) | Documentation, forums, service health dashboard | — | Core checks only |
| Developer | Cloud Support Associates, business hours, email | < 12 h general guidance | Core checks only |
| Business | Cloud Support Engineers, 24×7, phone/chat/email | < 1 h production down | Full set of checks |
| Enterprise | Adds a Technical Account Manager | < 15 min business-critical down | Full set of checks |
AWS has announced that the Developer, Business and Enterprise On-Ramp plans end on 1 January 2027, replaced by Business Support+, Enterprise Support and Unified Operations. Older study material will not mention this. The examinable principle is unchanged: the full set of Trusted Advisor checks requires a paid business-level plan or higher.
Key takeaways
- ✅Cloud computing is on-demand IT with pay-as-you-go pricing; the big change is economic, not technical.
- ✅The six advantages: variable expense, economies of scale, stop guessing capacity, speed, stop running data centres, go global in minutes.
- ✅IaaS you manage the OS; PaaS you manage the app; SaaS you manage only your data.
- ✅Three bill drivers: compute, storage, outbound data transfer.
- ✅On-Demand for unpredictable, Reserved or Savings Plans for steady, Spot for interruptible.
- ✅TCO must include facilities, staff, licences, disaster recovery and — above all — idle capacity.
- ✅Organizations gives consolidated billing and SCPs; SCPs only ever restrict, never grant.
- ✅Set a budget and a billing alarm on day one. The Free Tier is not a spending cap.