Serverless in AWS | AWS Lambda

Serverless in AWS – Overview

Serverless computing lets you build and run applications without managing servers. AWS handles the infrastructure, scaling, and high availability.

What is AWS Lambda?

AWS Lambda is a compute service that runs your code in response to events, automatically scaling and charging only for usage.

  • No server provisioning
  • Supports multiple languages (Python, Java, Node.js, Go, .NET, Ruby)
  • Common triggers: API Gateway, S3, SQS, DynamoDB, EventBridge

AWS Lambda Pricing – Example

You pay for:

  1. Requests: First 1M/month free, $0.20 per additional million
  2. Duration: $0.00001667 per GB-second

Example:

  1. Memory: 512 MB (0.5 GB)
  2. Execution time: 1 second
  3. Invocations: 1 million/month

Cost = (0.5 GB * 1 sec * 1M) = 500,000 GB-seconds
500,000 × $0.00001667 = $8.34

AWS Lambda Limits (Per Region)

Limit Default
Max function timeout 15 minutes (900 sec)
Max memory allocation 10 GB
Max package size (zipped) 50 MB (direct), 250 MB (with layers)
Concurrency limit 1,000 (can be increased)
Environment variables 4 KB
Ephemeral /tmp storage 512 MB (default), up to 10 GB
Execution role session duration 1 hour
SnapStart supported only in Java 11 and Java 17

Lambda SnapStart

  • Only for Java (11/17)
  • Snapshots the initialized execution state and reuses it across invocations
  • Reduces cold start time dramatically
  • Best for APIs and high-performance Java workloads

Customization at the Edge

Used with Amazon CloudFront to run code at AWS Edge locations.

CloudFront Functions

  • Ultra-lightweight JavaScript at the edge
  • Use cases: URL rewrites, header manipulation, redirects
  • Runs before cache at the viewer request phase
  • Lightning fast (<1ms)

Lambda@Edge

  • Full Lambda functions run at edge locations
  • Supports Node.js and Python
  • Runs at:
    • Viewer Request
    • Viewer Response
    • Origin Request
    • Origin Response

Ideal for authentication, A/B testing, user-based content delivery

CloudFront Functions vs Lambda@Edge

Feature CloudFront Functions Lambda@Edge
Languages JavaScript Node.js, Python
Use Case Lightweight logic Complex processing
Runtime Limit 1ms Up to 5 seconds
Size Limit 10 KB 50 MB
Execution Location Viewer Request only All 4 edge phases
Latency & Cost Lower Higher

Lambda by Default

  • Runs in a shared VPC
  • Cannot access RDS, EC2, or private resources unless configured
  • Internet access: ✅ by default
  • VPC access: ❌ unless explicitly configured

Lambda in a VPC

  • Required when Lambda needs access to:
    • RDS databases
    • Private subnets
    • Redis/Memcached (ElastiCache)

🔧 Needs:

  • VPC configuration (subnet + security group)
  • NAT Gateway if outbound internet access is needed

Lambda with RDS Proxy

  • Improves connection management to RDS (especially MySQL/PostgreSQL)
  • Benefits:
    • Efficient connection pooling
    • Helps avoid “too many connections” errors
    • Better performance under load

Invoking Lambda from RDS & Aurora

You can invoke Lambda from the database using:

  • Amazon Aurora (MySQL/PostgreSQL) via native SQL function aws_lambda.invoke()
  • Use cases:
    • Real-time triggers
    • Stream-to-Lambda data processing
    • Audit logging, notifications

RDS Event Notifications

Use Amazon SNS or Lambda to react to RDS events, such as:

Event Type Example
Availability DB instance restart
Security Credential changes
Backup Snapshot completion
Maintenance OS or DB patching

Subscribe via:

  • SNS topics
  • Lambda functions for automation
  • EventBridge rules for routing

Summary

Topic Summary
AWS Lambda Run code without servers
Pricing Pay-per-use (requests + duration)
Limits 15 min timeout, 10 GB memory
SnapStart Fast startup for Java
CloudFront Functions JS logic at edge (lightweight)
Lambda@Edge Full Lambda power at edge
Lambda in VPC Needed for private resource access
RDS Proxy Efficient Lambda DB connection mgmt
Invoke Lambda from RDS Aurora trigger to Lambda
RDS Notifications Event-based actions via SNS/Lambda

 

Back to blog

Leave a comment