Putting objects to work on a massive scale

Abstract

Distributed computing, and distributed object computing in particular, holds remarkable promise for future ISs and for more productive collaboration between our vast legacy IS base world-wide. This claim is not new to those who have read trade, research, or vendor literature over the past five years. GTE has made a significant attempt to benefit from this technology. We have found that it is currently considerably more difficult and less beneficial than the literature or its proponents would have had us believe. This chapter outlines challenges that we and others faced in attempting to put objects to work on a massive scale. The challenges were confirmed in a world-wide survey that we conducted of over 100 corporations that are attempting to deploy distributed object computing applications based on technologies such as CORBA, DCE, OLE/COM, distributed DBMSs, TP monitors, workflow management systems, and proprietary technologies.

Distributed object computing has offered a vision, significant challenges, some progress toward a computing infrastructure, and some benefits. Whereas distributed computing infrastructure and its interoperability is critical, application interoperability is the fundamental challenge to users of distributed computing technology. Large corporations spend on the order of $1 billion US annually addressing application interoperability. Although application interoperability is claimed to be the objective of distributed computing infrastructures, there has been little progress toward this critical ultimate requirement.

This chapter presents a view of distributed object computing from the vantage point of a large organization attempting to deploy it in the large scale. Requirements are presented in a distributed computing framework which is necessarily more comprehensive than anything offered by the distributed object computing vendors and proponents. It has four parts:

• Next-Generation Information Systems

• Computing Environment

• Computational Model

• Domain Orientation

Relative to this framework, we outline GTE's approach to distributed object computing, challenges we face and faced, why it is so hard, alternative distributed object computing infrastructure technologies, and an estimation of the state of these technologies. We conclude with the basic requirement for industrial-strength, enterprise-wide interoperable “applications.” This nontechnical requirement has always been a fundamental challenge for software.

No, Virginia, there is no distributed object computing, yet.