Reliable Actors with Retry Orchestration

Cloud developers have to build applications that are resilient failures and interruptions. We advocate for a fault-tolerant programming model the cloud based on actors, retry orchestration, tail calls. This builds upon persistent data stores message queues readily available cloud. Retry orchestration not only guarantees (1) failed actor invocations will be retried but also (2) completed never repeated (3) it preserves strict happen-before relationship across within call stacks. Tail calls can break complex tasks into simple steps minimize re-execution during recovery. review key application patterns failure scenarios. formalize process calculus precisely capture mechanisms of fault tolerance in this model. briefly describe our implementation. Using an inspired by typical enterprise scenario, we validate functional correctness implementation assess impact preparedness recovery performance.