Nile's Distributed Computing Site Architecture

The goal of the Nile project is to manage distributed computing resources, making an arbitrarily large cluster of commodity computers appear to the user as a seamless uniprocessor environment. This paper describes Nile's software architecture. 1 The Nile Project The Nile project 2,4] is building a distributed computing solution for CLEO III 3]. The goal is to provide a self-managing, fault-tolerant, heterogeneous system of hundreds of commodity workstations, with access to a distributed database in excess of 100 TB. Nile will allow any resource to be used transparently by any member of the collaboration, from anywhere within the collaboration. Nile must outlive its development phase, adapt to and scale with changes in CLEO's computing needs, be easily maintained, and be able to incorporate new software components as they become available; this led us to a distributed architecture based on CORBA. To program replicated objects we are using Electra 5], an ORB built on top of a reliable communication substrate; currently, Electra has adaptors for Isis 1] and Horus 7]. 2 Nile Architecture In Nile's software architecture, each of CLEO's collaborating institutions is a site, with a local set of users, and computational and data resources that process jobs submitted by local users or other Nile sites. The preference is to use local resources only, but the great disparity in sites' resources will require