Description
A very large distributed system (VLDS) is one based on a fast wide-area network connecting numerous organizations and individuals. The design of a VLDS involves problems and issues not present for smaller systems. These issues are centered in the areas of naming, communication paradigms and architectures, security, and kernel architecture.
DASH is a research project aimed at investigating VLDS design issues. The goals of the DASH project are 1) to predict the advances in computer and network hardware and application software that apply to VLDS; 2) to propose a set of design principles for VLDS, and 3) to experimentally validate these principles by building and testing a prototype VLDS kernel.
This report is a high-level view of the DASH project as of December 1986. After summarizing our ideas about the potential uses of VLDS's and our assumptions about the environment in which they will exist, we examine several design areas. In each area, we offer some possibly controversial assertions, attempt to justify these assertions, and describe how the assertions have guided the design of DASH.