Serverless Development on AWS • Sheen Brisals, Luke Hedger & Khawaja Shams • GOTO 2024

Serverless adoption requires understanding its characteristics and capabilities before diving too deep - it should be viewed as a tool rather than just a methodology or philosophy

Cost management in serverless differs from traditional infrastructure - it follows a consumption-based model where costs scale with actual usage rather than fixed infrastructure costs

Enterprise adoption barriers include lack of educational materials, misconceptions about serverless being only for “toy projects”, and resistance due to unfamiliarity with cloud technologies

Common myths that need addressing include security concerns, local development capabilities, and testing approaches - these should be viewed as opportunities rather than blockers

Real enterprise case studies (Taco Bell, PostNL) demonstrate serverless is viable for critical production workloads and can handle spiky traffic patterns effectively

When evaluating serverless, start with lower-criticality production workloads to test and learn before expanding adoption

Continuous refactoring is important as cloud technologies evolve - architecture should be allowed to evolve rather than remain static

Serverless-first approach doesn’t mean serverless-only - it’s important to assess which workloads are appropriate for serverless architecture

Team education and establishing proper development processes/methodologies are critical for sustainable serverless adoption

Infrastructure as code and managed services integration are key components of successful serverless implementations