Synchronize your watches: Part I: General-purpose solutions for distributed real-time control

In many distributed systems that involve interactions between physical objects, human beings and environment, the precise time of an event is very important. This is particularly true for automated control applications, wherein the knowledge of the exact time of occurrence of a certain event, or the ability to enforce a given action at a precise instant, is crucial for the correct operation of the entire system (or, at least, for its accuracy and performance). Typical examples are the execution of transactions in enterprise resource planning (ERP) systems or the coordination of movements in multi-axis, numerically controlled machines. The required degree of accuracy depends on the specific application considered. For the two cases mentioned above, for instance, it is in the order of several seconds and fractions of microseconds, respectively. It is worth noting that even in the case of distributed applications that have planning horizons of days or even weeks, such as ERP, the time resolution on different machines (as provided by the underlying operating systems) related to, e.g., the last-access/modification time for files, is typically 1 s or less.