Distributed Object Management

A system of distributed objects is a set of processes not sharing memory where processes can invoke the methods of objects owned by other processes in the set, just as if those objects where owned by the caller (it must provide some degree of transparency).

The management of distributed objects consists of the mechanisms employed by the system for implementing remote references, remote method invocations and any other feature provided by the system (like distributed garbage collection).

Objectives of the course

Knowing what are the trade offs of the designing of distributed objects systems from an implementation point of view.
Understanding the implementation of such systems.
To be trained for writing or modifying the code of a simple system of distributed objects.

Required previous knowledges

Programming in Java with RMI or CORBA.

Introduction:
Network Objects for Java:
The implementation of network objects for Java.
Strategies for distributed garbage collection

Bibliography

For a deepper discussion of the concepts seen in this course, I recommend reading the following papers:

[RMI'99] Sun Microsystems, Java Remote Method Invocation(RMI), http://java.sun.com/j2se/1.3/docs/guide/rmi/index.html, 1999.
This document serves to study a genuine example of a distributed object system.
[Birrel93a] Andrew Birrell, Greg Nelson, Susan Owicki and Edward Wobber, Network Objects, SRC Research Report 115, Digital System Research Center, 1993.
This paper shows a simple way to implement distributed object systems such as Java RMI. Moreover, I suspect that Java RMI is based entirely on this paper. The only problem is that all the code is written in Modula-3 which is unknown for most people. But don't be affraid about that.
[Birrell93b] Andrew Birrell, David Evers, Greg Nelson, Susan Owicki, and Edward Wobber, Distributed Garbage Collection for Network Objects, SRC Research Report 116, Digital System Research Center, 1993.

Download this document in zipped html format.