The Virtual Computing Environment

A network of supercomputers and high-performance workstations appears to be the only reasonable way to provide adequate computing resources for the Grand Challenge problems of the next century. Such a collection of computers and supporting software environments is called a virtual computing environment (VCE). This paper describes the motivation and goals of the VCE project, followed by a description of the system. The paper concentrates on the runtime aspects of the VCE, and concludes with a discussion of a small prototype system that has been built using the Isis distributed toolkit. Advances in software technology have produced tools and environments for the development of parallel and distributed applications. These advances have resulted in the proliferation of a large number of diier-ent architectural classes like SIMD computers, MIMD computers, vector computers, etc., where each class represents a set of diierent trade-oos in design decisions. Each architectural class is tuned to deliver maximum performance to a speciic set of applications, however, none of the existing computer systems are general enough to address all classes of applications and provide the desired performance levels. Applications that require the resources of a super-computer are quickly reaching the point where it will be diicult to nd a single computing platform on which they may be executed. The grand challenge problems of 1995 and beyond will require a memory capacity of 1000GB and system speeds of 1 TTop 6]. Given the current pace of technological advancement it is unlikely that a single supercomputer will be able to provide the resources required for these problems. A network of supercomputers and high performance workstations appears to be the only reasonable way to provide adequate computing resources for these large applications. Programmers developing parallel applications are faced with the daunting task of selecting the correct architecture, nding the appropriate software tools, and must handle many low-level issues such as load balancing. The situation today is similar to that of the pre-compiler era of computing. Before the rst FOR-TRAN compilers were available, programmers could nd the tools required to build their applications, however they were forced to program in assembly language and had to deal with low-level machine issues in addition to the task of developing their programs. The rst high-level languages made programming easier by providing an abstract model of computing which hid the details of the low-level machine and enabled the programmer to concentrate on the application. The same …