Research Seminar - September 03, 1999

Abstract

While these ideas were closer to wishful thinking 3-4 years ago, their widespread usage is fast becoming a reality. The idea of distributed objects is as old as distributed computing itself. However, it simply cannot be put into general use before the arrival of Java. Early distributed object research began with the development of the Smalltalk Model-View-Controller idea. In fact many of current ideas were formulated at that time. Unfortunately, nothing short of a software revolution can provide sufficient momentum for wide spread use of distributed computation.

CORBA is the first standard for distributed objects. Because it was designed to support all languages, it is extremely complex and its objects are fairly immobile. Java Remote Method Invocation (RMI) was introduced by SUN as a Java API for distributed objects. When distributed objects are integrated with a single language, many distinct advantages appear. They include write-once-run-anywhere, dynamic type checking, interface verification, standardised object serialisation, reference management, garbage collection, and security. RMI also provides a rudimentary mechanism for object migration. Unfortunately, RMI remains more like a research tool than a commercial product. It is complex for users. About 2 years ago, the first ObjectSpace Voyager was released. It is a Java based distributed ORB for migrating objects. Object migration opens up many computing issues both in the architecture of system as well as implementation design.

In this talk, a classification model for various migrating objects will be given. Current implementations will be discussed. Demonstration will be given on some examples. Lastly, I will expend on the exciting implications of migrating objects in computing: hardware design, software components, reflection, activation, persistent objects, databases, browser technology, electronic commerce, computer communications, securities, system organisation and computer system architecture .

p.s. I will also discuss the hindrances to the progress in migrating objects and why current systems are not desirable bases for this ambition in distributed systems. We need a revolution!