Design of the Kan distributed object system

Distributed software problems are often addressed with object-oriented solutions. Objects provide the benefits of encapsulation and abstraction that have proven useful in managing the complexity of sequential code. However, the management of distributed objects is typically by means of complex APIs, such as CORBA, DCOM, or Java RMI. The complexity of the APIs is itself a hurdle to the writing of efficient, robust programs. An alternate approach is to provide the programmer with a simple interface to an underlying object management layer that provides efficient access to objects and sufficient power for common distributed programming tasks. This paper describes the implementation of the Kan system. It has a clear, simple object model with powerful semantics, embodying such concepts as atomic transactions, asynchronous method calls, and multithreading. The model primitives help the programmer avoid common concurrent programming errors, allowing clean expressions of concurrent algorithms. Kan provides distributed objects (i.e., objects that can migrate or be replicated), rather than the remote objects of Java RMI. Nevertheless, Kan optimizations provide runtime object accesses that are as efficient as or more efficient than accesses made to a similar distribution control layer over Java RMI. We describe the optimizations and measure their runtime impacts.