ConTracts: a means for improving reliability in distributed computing

ConTracts are designed to build large distributed applications from short ACID-transactions. They provide a framework for treating the resulting computations as recoverable objects. The key idea is to use an explicit control flow description by the system to reliably control and manage a long-living application´s execution. No system failure at execution time should trouble the application programmer. Even in the case of a node crash an affected application should not be aborted but should be continued in the forward direction.<>