Abstract

In recent years high performance computing (HPC) sites have been established in various countries. In Canada connectivity between HPC sites is provided by CA*net 2, Canada’s high performance network (HPN). Users of HPC facilities currently submit jobs as, typically, program source files requiring local compilation for execution within the site and all computational resources are restricted to single site usage. Increasingly, it is the case that resource requirements extend beyond the capabilities of single HPC sites. Requirements may include more CPU cycles, access to files, user interaction and executing programs in distributed collaboration scenarios and with both present Internet and HPN based communication occurring among HPC sites. Local site policies often inhibit or disallow direct exchange of data and programs for security reasons. To solve such problems it is necessary to establish appropriate policies and protocols to enable communication between distributed processes. It is imperative to reach consensus on the issues and practices of effective, secure sharing policies among HPC sites in order to develop appropriate policies for networked resource sharing. In this paper we provide several examples to illustrate the types of computational and networking requirements which might typify inter-networking for shared distributed computing. Additionally, we summarize various points and opinions expressed by audience discussants as part of this panel presentation.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call