AWS Compute Services
Identify AWS compute services
What you will learn
- Distinguish EC2 from Lambda in terms of who manages the server and the billing model.
- Tell ECS apart from EKS as two different container orchestration services, and see where Fargate fits.
- Explain how Auto Scaling works with a load balancer to absorb rising demand automatically.
- Pick the right compute service for a given scenario instead of just memorizing names.
AWS compute is a spectrum, not a single point. At one end is Amazon EC2: virtual servers you fully control — you pick the OS, install whatever you want, and manage it like your own machine. At the other end is AWS Lambda: you write only a function, and AWS runs it on an event without you ever seeing a server. Containers sit in between, giving finer control than Lambda without full server management like EC2.
The core services
| Amazon EC2 | Virtual servers with full control — you choose the instance type and OS, and manage patching yourself. |
|---|---|
| AWS Lambda | Serverless compute: you write a function triggered by an event, and pay only for execution time. |
| Amazon ECS | AWS's own fully-owned container orchestration service — simpler to configure, tied to the AWS ecosystem. |
| Amazon EKS | AWS-managed Kubernetes — for teams already using Kubernetes and wanting standard compatibility with it. |
| AWS Fargate | A serverless compute layer for containers, usable with either ECS or EKS — you never manage the EC2 instances behind the containers. |
EC2 vs. Lambda
| Amazon EC2 | AWS Lambda | |
|---|---|---|
| Server management | Fully managed by the customer | No visible server at all |
| Billing | For however long the instance runs, used or not | Only for the function's actual execution time |
| Best suited for | Continuous, predictable loads needing fine environment control | Intermittent, event-triggered loads (a file upload, an API request) |
How Auto Scaling works with a load balancer
An automatic response to rising demand with no human intervention.
Request load rises
CloudWatch detects a defined metric threshold crossed
Auto Scaling launches new EC2 instances
The load balancer distributes requests across every healthy instance
What matters on the exam
- You don't need every compute pricing formula here — objective 4.1 covers pricing in more depth. Here what matters is knowing which service fits which scenario.
- "Intermittent, unpredictable load" in a question's wording usually points to Lambda, not EC2.
ECS is not EKS
Both orchestrate containers, but they are not the same thing: ECS is an orchestration system AWS owns exclusively, while EKS is a managed version of open-source Kubernetes. A question that explicitly mentions "Kubernetes" is pointing to EKS, not ECS.
Quick check
Answer in your head first, then reveal.
Which compute service bills only for execution time, not uptime?Reveal the answer
AWS Lambda.
What is the difference between ECS and EKS?Reveal the answer
ECS is AWS's own container orchestration; EKS is a managed version of Kubernetes.
What exactly does Fargate do?Reveal the answer
It runs containers (via ECS or EKS) without managing the EC2 instances behind them.
What does CloudWatch do in the Auto Scaling flow?Reveal the answer
It monitors metrics and fires the alarm that triggers Auto Scaling.
Sign in to track your progress on this topic.
Your next step
Read the lesson, then mark it complete
Sources
Used to verify the facts. The writing is original to Passuit.