title: Forgetting DCE

by: Scott Bradner

Some of you might remember Distributed Computing Environment (DCE) but it's not clear that some of the industry pundits or venture capitalists do. Or at least they haven't internalized a principal reason that DCE is, to put it politely, not prevalent today.

For those who do not remember, Distributed Computing Environment is a set of technologies developed by the Open Software Foundation (now called "The Open Group" www.opengroup.org) that enables a computer user to make use of network-based resources to augment their local computer. DCE, quoting from an IBM web page 'is a comprehensive suite of integrated, yet modular, products which support transparent file access and secure resource sharing in heterogeneous, networked computing environments." (For more info on DCE see http://www.faqs.org/faqs/dce/faq/.)

I'm sure the Open Group will call this simplistic, but in my mind a major reason that DCE was developed was to share resources, such as disk and processor cycles, over a network because having enough dedicated resources for individuals was too expensive. With DCE the user can get access to databases without having to have a local copy and can get heavy-duty processing done without having to have as powerful a computer on their desk.

But the DCE proponents did not take into account the continued development of technology. Before the DCE specifications could be fully developed disk and computer technology developed enough to negate much of the assumed advantages of using DCE. DCE was based on the assumption that the cost of management of the use of distributed resources would remain less than the cost of replicating those resources. This assumption did not prove to be long lived in the case of DCE.

There just may be a lesson in the history of DCE for those who are considering investing in peer-to-peer networking, storage as a service offering or maybe even virtual private networks (VPNs).

I am leaving out a number of other arguments that were made in the case of DCE, single signon, centralized backup, centralized authorization management, and more. Some of these arguments are now made for the newer technologies -- they may prove to be as non-decisive as they did for DCE. I am also leaving out the ego factor that leads network managers to think that they should control everything that connects to their networks. That factor is harder to analyze -- some of the egos are rather strong.

An undercurrent of Clayton Christensen's "The Innovator's Dilemma" is that it is quite hard for people to take into account the fact that technology does not stand still when evaluating their options. It is much too easy to see what you can buy today and assume that it represents what will be available in the future. A case of this may be the pundits that dismiss using the best-effort Internet for telephony - all they can see is that it would not work well enough for them today. They forget that using today as a guide led to developing DCE.

disclaimer: Harvard knows yesterday and today rather well but has trouble with tomorrow, as do I, but I provide the above caution anyway.