Title

AWS re:Invent 2023 - Refactoring to serverless (SVS305)

Summary

Serverless is not just about infrastructure management but also about reducing application code through automation.
The session focused on making existing serverless applications more serverless-native, not on the six Rs of cloud migration.
AWS CDK allows infrastructure as code using familiar programming languages and provides higher-level abstractions for more compact and readable code.
Cloud automation spans provisioning, deployment, composition, and configuration, with composition and configuration being crucial in serverless.
Refactoring to serverless involves using serverless features like Lambda Destinations and EventBridge Pipes to separate business logic from application topology.
The session presented examples of refactoring, including replacing choreographed Lambda functions with Step Functions for better workflow visibility and transactional integrity.
Refactoring is driven by leveraging platform capabilities, incorporating new AWS features, improving runtime characteristics, and managing costs.
Operational aspects like latency and cost were discussed, with the conclusion that refactoring can often lead to cost savings and improved performance.
Team structures and tooling play a role in refactoring, with the need for continuous refactoring as part of software delivery.

Serverless refactoring can lead to more robust and scalable applications by replacing application code with automation code.
AWS CDK is a powerful tool that allows developers to use familiar programming languages to define infrastructure and application architecture as code.
The importance of composition and configuration in serverless applications is highlighted, as infrastructure provisioning and deployment are simplified in a serverless environment.
Refactoring patterns can be categorized into leveraging platform capabilities, evolving with AWS features, and improving runtime characteristics like latency, availability, and cost.
Operational considerations such as latency and cost are not necessarily negatively impacted by refactoring; in fact, they can benefit from it.
Team structures should allow for the development team to handle both application code and automation code to ensure efficient refactoring practices.
Continuous refactoring should be integrated into the software delivery process, and it can be a driver for improving development practices and reducing complexity.
The session encourages a mindset shift towards seeing infrastructure and application code as part of a continuum, where refactoring can move functionality from one to the other seamlessly.