Network Access Capability to Systems as A Factor in Distributed and Cluster Computing

Abstract

This paper reports on perspectives concerning the importance of network access as the interconnecting fabric in distributed and cluster computing. It has relevant application to the areas of grid computing, as well as for application delivery requiring end-to-end, specific network capability for the successful delivery of applications between two or more networked environments. The complexity of secure operating environments, packet and cell switching and processing capabilities, optical capacity, networking overhead, and internetworking standards all have the potential to impact on the capability of the interconnecting network fabric. This work examines strategies for high speed optical networking and attempts to create a framework for evaluation of the access to system resources that are envisioned to be required in supporting the access requirements of a distributed HPCS computing environment. Network interfaces, network latency, protocols and connection management strategies in hybrid network infrastructure environments may result in a synergy of latency issues. This synergy must be examined to determine their impact on networking environment, in addition to effects at the processing layer for job scheduling and management of system resources.