Reliability: Ability to recover from infra and app issues

Adapt to changing demands in load

Best Practices

Automate recovery from failure

Health checks and Auto scaling

Managed services like RDS can automatically switch to standby

Scale horizontally (Reduces impact of single failure)

Maintain Redundancy

Multiple Direct Connect connections

Multiple Regions and Availability Zones

Prefer serverless architectures

Prefer loosely coupled architectures: SQS, SNS

Adhere to Distributed System Best Practices

Use Amazon API Gateway for throttling requests

AWS SDK provides retry with exponential backof