There also is no good mchanism for doing distributed computing in any simple way across systems. For example, if I want to invoke a compile build across systems under Windows, how is this done? Or, if I'd like to take 10 UNIX stations, 5 Windows systems and 4 Macs and turn this into a generic processor farm. There are tools for doing similar things, but there is no simple process for setting this up under one system, much less several. There is also no standard interface for such a need, especially when you want to start sharing files and communicating between the remote processes.
Even in systems that deliver feature sets close to those described here, there is no easy security built in. Or, where there are security measures, it requires a huge administrative overhead. There is no easy way to do any of this. One motivation, therefore, was the desire to have an open environment in which all this and more is easily accessable, and not just to expert programmers and all-powerful sysadmins, but for the average user as well.
A second, but equally important motivation was to study such systems. By building a model of an open, distributed system, which contains all of our core feature requirements, we could better understand the complexities and issues in building a full scale version, and try to understand why people have designed other similar systems as they have. As primarily a research project, instead of an attempt to build a software package, we could take the time to really understand the security, performance, and portability issues, as well as the design choices that need to be made. At this point we have produced not only a demoable model, but also a description of the problems and some of the sollutions in constructing such a system.