Durable Executions, Defined (journal.resonatehq.io)

A recent post has brought clarity to the concept of "durable executions," which has gained traction as large language models (LLMs) evolve from interactive chatbots to continuous agents engaged in complex workflows. The article rigorously defines durable executions as "Interruption-Agnostic Definitions of Functions that result in Interruption-Tolerant Executions of Functions." This clarification emphasizes the distinction between what constitutes “normal code” (code that is not designed with interruptions in mind) and “fault tolerance” (the ability to handle interruptions during execution). This definition is significant for the AI/ML community, as it addresses the ambiguity surrounding various existing frameworks that claim to provide fault tolerance without a clear understanding of how to achieve it. Resonate's Distributed Async Await framework aims to simplify durable executions, moving away from proprietary DSLs and complex workflows. Instead, it promotes a minimalistic approach that allows developers to write familiar code while ensuring the system can handle interruptions gracefully—demonstrating that developers can maintain control over their programs’ behavior without extensive restructuring or added complexity.