A team wants to run containers on AWS. How do you decide between ECS, EKS, and Fargate?
What they are really testing: Whether you understand that these are not all the same kind of choice, orchestrator versus compute model, and whether you weigh operational overhead against portability and ecosystem instead of just picking the trendiest option.
A real interview question
A team wants to run containers on AWS. How do you decide between ECS, EKS, and Fargate?
What most people say
drag me
“I would use EKS because Kubernetes is the industry standard and everyone uses it.”
It picks on hype, not fit, and treats three different choices as one. It ignores that EKS carries real operational overhead, never mentions Fargate as a compute model, and gives no reason tied to the team skills or workload.
The follow-ups they ask next
The team chose Fargate and now the bill is higher than expected at steady high load. Why, and what would you change?
Fargate prices the convenience of no node management; at steady high utilization, well-packed EC2 nodes (and Savings Plans/Spot) are usually cheaper. Move steady baseline workloads to EC2 capacity and keep Fargate for spiky or low-utilization tasks.
When does the operational cost of EKS actually pay off over ECS?
When you genuinely need Kubernetes: multi-cloud portability, the ecosystem (operators, Helm charts, CRDs), or a team already fluent in k8s. If none of those apply, EKS is paying complexity for features you will not use.
What the interviewer is listening for
- Separates orchestrator from compute model
- Defaults to the lowest-overhead option that fits
- Justifies EKS with portability or ecosystem, not hype
- Knows Fargate vs EC2 is a cost-vs-control trade-off
What sinks the answer
- Picks Kubernetes because it is trendy
- Treats ECS, EKS, and Fargate as one choice
- Ignores operational overhead of EKS
- No reasoning tied to team skills or utilization
If you genuinely do not know
Say this instead of freezing. Reasoning out loud from what you do know beats silence every single time, and a good interviewer is listening for exactly that.
“These are two decisions: [orchestrator: ECS vs EKS] and [compute: Fargate vs EC2]. ECS is [AWS-native, lower overhead]; EKS is [managed Kubernetes for portability/ecosystem but more complexity]. Fargate is [serverless, no nodes]; EC2 is [more control, cheaper at steady high load]. I default to [ECS on Fargate] and move to [EKS for portability or EC2 for steady high utilization].”
Keep going with containers
Mid
A service is suddenly returning 500s in production. Walk me through how you would troubleshoot it.
Mid
Design durable, highly available storage for user-uploaded images. Which service do you choose and why?
Mid
An Azure VM cannot connect to an Azure SQL Database. Walk me through how you debug the networking.
Mid
How would you manage application secrets (DB passwords, API keys) across environments? Walk me through the options and their trade-offs.
Mid
Your team provisions infrastructure with code. Why does idempotency matter, and what breaks without it?
Mid
Compare blue-green and canary deployments. When would you reach for each?
Knowing the answer is not the same as recalling it under pressure
Sign in to send the questions you fumble to spaced recall, so they come back right before you would forget them, and learn the concepts behind them with hands-on labs.
Start free