Workshop on Scalable, Uninterruptible Computing
  The goal of this half-day workshop is to provide a forum for researchers and practitioners, from academia and industry, to discuss the main challenges and practical solutions for implementing scalable uninterruptible computing systems.

Today's networked systems employ a highly heterogeneous collection of COTS, such as computers, routers, switches and storage. Clusters of advanced workstations, built from COTS, already play an important role in many industrial and military applications. Their behavior is mostly dynamic, as they are expected to undergo continuous changes in response to unpredictable events such as addition/deletion of computational resources, upgrades on hardware/software capabilities and reconfiguration. As the complexity of these systems is growing at a very explosive rate, efficient tools and techniques must be developed in a cohesive framework.

A comprehensive, multi-disciplinary approach is required to the problems associated with scalable networked systems with the ultimate objective to provide uninterruptible computing and high quality of service. Novel architectures need to be developed for providing the basis for efficient, scalable and reliable computation and communication. The architecture should incorporate system entities at different functional levels thus being able to account for complex interactions. This will require a detailed characterization of a hierarchical environment which closely monitors behavior of all resources (hardware and software). Uninterruptible computing requires the composition of many entities (for example routing, fault tolerance, reconfiguration and recovery-based computing). The approach is multi-disciplinary, because it utilizes software with a clear understanding of the underlying hardware resources (routers, switches, network-interface controllers) by analytical and algorithmic modeling techniques. This approach will allow to build scalable networked-based systems, based on COTS, that are inexpensive, accessible, scalable and dependable. These systems will insure transparent fault tolerance and uninterruptible computing.


Dimiter Avresky
(Northeastern Univ., USA)

Program Committee

D. Avresky (NEU, USA)
J. Hayes (Univ. Michigan, USA)
B. Johnson (Univ. of Virginia, USA)
F. Lombardi (NEU, USA)
R. Maxion (CMU, USA)
J. Plank (U. of Tennessee, USA)