Extending distributed real-time java with Remote Memory Areas

Current approaches towards distributed real-time Java are mainly based on the idea of having user-defined remote objects, allocated in servers that may be invoked from clients. This article extends this support included in real-time Java with an extension called Remote Memory Areas (RMAs). RMAs offer a method that allows running user-defined code in a generic server that may be reused by several applications (i.e., it does not require defining one type of remote sever per application). The paper describes the abstraction, which is backward compatible with main approaches in distributed real-time Java, and provides empirical evidence on its performance on a use-case application.