AWS Serverless Application Development

Source: My personal notes from AWS training courses and cloud training at work

Best Practices in Serverless

Overview:

• Serverless design
• Functions
• Infrastructure as Code
• Testing
• Development Security Operations (DevSecOps)

What is Serverless?

• A culture with focus on business value using technology that is suitable for building and running applications

• Benefits:

  • Faster development to production to give business value
  • Continuous experimentation

• Example service:

  • Design: Event Source –> Function –> Call other services
    • Events like data state changes, requests
  • AWS Lambda service handles Function part
    • Functions can transform data

About AWS Lambda

3 phases

• Init
  • Create new environment to run execution
  • Download code/image
    • Packaged as layer
  • Initialize runtime
  • Run function pre-handler initialization: can be optimized by user
    • Example: Get secrets, handle connections, peristent data
• Invoke
  • Run function handler code: can be optimized by user
    • Good practice to move business logic to a separate area inside code of function
  • Continue with more requests
• Shutdown
  • Runtime turns off

Design

Considerations on using functions:

• How much logic to put into functions?
  • Avoid large functions
  • Can separate complex functions to sub functions, like routing logic
    • Individual actions can be split into its own function or functional groups
    • Balance number of functions with management, security considerations
    • Can group functions by:
      • Business context
      • Teams
      • Security permissions
      • Technology
• How much is the function doing?
  • Business logic
  • Complex logic
  • Is it too simple?

Example architecture in AWS

S3 front end, API management, managed services for queue/workflow/pubsub, AWS Lambda for functions, S3 storage

Client –> API Gateway –> Multiple functions –> Database / Backend services

• Remember API gateway can integrate directly with backend services if it makes sense instead of using functions as only as simple proxies
• Orchestration (central control): AWS Step Functions
  • Example: write steps which are functions, coordinate steps between functions using low code
    • Intrinsic functions in step functions: arrays, JSON, encoding/decoding, math and other
• Choreography: Amazon EventBridge
  • Sources –> EventBridge (event is pushed to target) –> Targets
• Example custom integration: Amazon DynamoDB with Change Data Capture (CDC) –> Function > Event handling like EventBridge

Monitoring and Performance Examples

• Use AWS X-Ray to check AWS Lambda function performance and execution information and run times
• Optimize logging: structured logs stored with S3
• SnapStart for Java functions - takes a shot of most of init phases for resuming snapshot for repeated calls
• Adjust AWS Lambda capacity, optimize memory using AWS Power Tuning and Lambda recommended settings after runs have been tracked
  • Running on Graviton2

Concurrency

• Concurrency is running multiple requests at the same time assuming AWS Lambda capacity is not reached
• Can reserve concurrency for given function

Infrastructure as Code

Create AWS Cloud Formation templates using simplified:

• AWS Servless Application Model (SAM)
  • Reuse policy template for security practices
• AWS Cloud Development Kit (CDK)

Testing

• Do testing to improve code speed and quality
• Help document code functionality
• Unit, integration, end to end tests
• Testing strategies for automated tested: Emulators, Mocking (mock cloud components locally), Testing in Cloud

DevSecOps

• Implement controls like: pipelines, private endpoints and networks, logging enabled with retention periods, resource tags, only use supported runtimes, environment variables are encrypted at rest
• Code/IAC linter, Config, certificates
• Security analysis