State, Data, and Consistency in Serverless Systems
摘要
State is where distributed systems become difficult. Messaging and serverless triggers can decouple work, but correctness depends on data. Once a state is spread across Functions, queues, databases, and external services, the simple assumptions of a monolith begin to fail. Reads can be stale, writes can conflict, and operations that appear to succeed independently can still leave the system in an inconsistent state.