aws-serverless

AWS Serverless - Production-Grade Serverless Application Development Skills

Skill Overview

The AWS Serverless skill provides complete patterns and practices for building production-grade serverless applications, covering Lambda function development, API Gateway integration, DynamoDB data layer, event-driven architecture, and SAM/CDK deployment options.

Applicable Scenarios

1. Building RESTful APIs and HTTP Services

When you need to quickly build and deploy API endpoints, you can use the API Gateway with Lambda integration pattern. The skill provides complete SAM template examples that show how to configure HTTP APIs, set CORS, and bind routes to specific Lambda function handlers.

2. Asynchronous Message Processing and Task Queues

For scenarios that require decoupling and reliable retries, the skill provides a full pattern for SQS-triggered Lambdas. It includes batch size configuration, partial failure handling (ReportBatchItemFailures), dead-letter queue setup, and other production practices to ensure high reliability of message processing.

3. Event-Driven Microservices Architecture

When building loosely coupled, scalable microservices, the skill demonstrates how to use services like SNS/SQS to implement event-driven patterns. It includes Lambda initialization best practices, context reuse, and how to avoid common performance pitfalls.

Core Features

1. Standard Lambda Handler Pattern

Provides standard Lambda function structures in both Node.js and Python, including:

Client connections initialized outside the handler (reused across invocations)

Event source parsing and error handling

API Gateway–compatible response formats

Structured logging

2. API Gateway Integration and Deployment

Complete SAM template examples covering:

HTTP API and REST API configurations

Path parameter and request body handling

IAM permission policies (DynamoDBReadPolicy, DynamoDBCrudPolicy)

Environment variables and CORS settings

3. SQS Event-Driven Processing

Production-grade message processing patterns, including:

Batch message processing and partial failure reporting

Visibility timeout configuration (recommended to be 6× the Lambda timeout)

Dead-letter queues and retry strategies

Error tracing at the message ID level

Frequently Asked Questions

How long is Lambda cold start typically? How to optimize?

Cold start time varies by runtime, memory configuration, and deployment package size, typically ranging from a few hundred milliseconds to a few seconds. Optimization measures include: keeping deployment packages lean, choosing appropriate memory allocation (512MB+), placing initialization logic outside the handler, and avoiding unnecessary dependencies. If a Lambda is inside a VPC, ENI creation will add additional latency.

How should I choose between SAM and CDK?

SAM (Serverless Application Model) is better suited for rapid development and simple scenarios, offering simplified syntax and common patterns. CDK (Cloud Development Kit) is more suitable for complex infrastructure and scenarios requiring programmatic abstractions, supporting languages like TypeScript and Python for infrastructure code. Both can deploy the same types of resources; the choice depends on team preference and project complexity.

Why do Lambdas in a VPC perform worse?

Lambdas in a VPC need to create Elastic Network Interfaces (ENIs), which increases initialization time. Recommendations: enable VPC only when access to VPC-resident resources (such as RDS or ElastiCache) is required; configure an adequate number of ENIs; if only accessing public services, keep Lambdas outside the VPC. The "Sharp Edges" section in the skill provides a detailed configuration checklist.