An Object Model for Flexible Distributed Systems

this paper we describe a new model for constructing operating systems and applications in an integrated fashion. Compared to current approaches we provide high-level primitives for supporting distributed and parallel applications. We also provide the flexibility to configure both applications and kernels to only include the functionality that is actually used. The model we describe is based on objects. Objects are used to structure both applications programs and operating system kernels. They also provide the application interface to the operating system kernel, and access to hardware devices for both kernels and applications. By providing structuring mechanisms for large (distributed) objects, we believe that applications will be are easier to build. At the same time we provide flexibility by allowing extensions of operating system kernels and applications with new objects at run time[8], and by providing a way to bind to objects dynamically. An important aspect of a distributed system is the scalability of the system [6]. A scalable system should not depend on centralized resources or on algorithms that need global information. At the same time, a flexible system can use different algorithms depending on the situation. For example, the use of broadcasting and multicasting on a local Ethernet can be quite effective but should be avoided on a world wide scale. In this paper we discuss an object model that provides two kinds of objects: local objects and distributed objects. In Section 2 we describe the nondistributed (local) objects, followed by distributed objects in Section 3. We compare our work to that of others in Section 4. 2 Local objects