Indiana University Bloomington

Luddy School of Informatics, Computing, and Engineering

Technical Report TR567:
Breaking the Barriers: High Performance Security for High Performance Computing

Kay Connelly and Andrew A. Chien
(Aug 2002), 7 pages pages
[A version of this report will appear in the New Security Paradigms Workshop, 2002.]
Abstract:
This paper attempts to reconcile the high performance community's requirement of high performance with the need for security, and reconcile some accepted security approaches with the performance constraints of high-performance networks. We propose a new paradigm and challenge existing practice. The new paradigm is that not all domains need long-term forward data confidentiality. In particular, we take a fresh look at security for the high-performance domain, focusing particularly on component-based applications. We discuss the security and performance requirements of this domain in order to elucidate both the constraints and opportunities. We challenge the existing practice of high-performance networks sending communication in plaintext. We propose a security mechanism and provide metrics for analyzing both the security and performance costs.

Available as: