A More Useful Programming Model

The potential failure cases in a distributed system are numerous and hard to predict. Modern software engineering practices rely on a programming model where errors are known in advance or are handled with blanket error handling policies. From the perspective of a software engineer, a more useful model would be one where it is possible to adapt to failures as they occur. In order to recover from failures as they occur it is necessary to reliably suspend code execution and state. One of the primary benefits of such a model would be greatly reduced cognitive overhead when building and maintaining distributed systems.