Basic Information
| Field | Details |
|---|---|
| English name | Serverless Architecture |
| Full name | Serverless Architecture |
| Chinese description | 无服务器架构 |
| Japanese description | サーバーレスアーキテクチャ |
| Exam frequency | ⭐⭐⭐⭐⭐ |
| Often confused with | 容器 / EC2 |
In one sentence
Serverless does not mean no servers; AWS manages server provisioning, maintenance, and scaling.
Typical components
- API Gateway or EventBridge accepts requests and events.
- Lambda runs short event-driven functions; Fargate runs containers without managing EC2.
- SQS / SNS buffers, decouples, or broadcasts; DynamoDB / S3 stores data.
- Step Functions manages workflows, while CloudWatch provides logs, metrics, and alarms.
Lambda, Fargate, and EC2
- Lambda: short, event-driven functions.
- Fargate: containers without EC2 management; ECS or EKS still orchestrates them.
- EC2: long-running processes, special OS needs, or maximum control.
Exam focus
- Lambda processing SQS is at least once, so functions should be idempotent with correct Visibility Timeout and DLQ settings.
- Serverless still requires IAM, networking, data, retries, concurrency, monitoring, and cost management.
- Automatic scaling does not guarantee downstream databases can handle the same concurrency.
- Choose compute from execution duration, runtime model, and control requirements.
Common misconceptions
- Serverless is neither free nor limited to Lambda.
- Fargate is serverless compute, not a container orchestrator.
- Lambda is not for long-running processes, and every container app need not become functions.
Key takeaway
Serverless shifts management responsibility: functions → Lambda, containers → Fargate, maximum control → EC2.
Related services
Lambda, Fargate, ECS, EKS, SQS, EventBridge, API Gateway, DynamoDB, CloudWatch.