John Dougherty

Split a system in the wrong place and you get a distributed monolith: coupled services with network hops instead of function calls. This session tackles the problem most teams get wrong, where to draw the boundary, using Event Storming.

Distributed workflows need coordination, and most teams pick one approach everywhere: a central orchestrator, or independent services reacting to events. Both fail at scale: orchestration bottlenecks every team's releases, and choreography lets flows stall silently because no component owns the outcome.

Master advanced patterns for building hybrid orchestration-choreography architectures from the ground up. Implement orchestration collectors using EventBridge and Step Functions for coordinated workflows, while creating choreographed microservices with Lambda and SQS for autonomous operations.

Cost optimization is crucial for scaling IoT workloads on AWS. This chalk talk explores design patterns and AWS tools that you can use to continuously optimize workloads, maximizing resource usage while limiting infrastructure costs.