Title: CaPaS: an optimal security-aware cache replacement algorithm for cluster storage systems

Authors: Mais Nijim; Xiao Qin; Muhittin Yilmaz

Addresses: Department of Electrical Engineering and Computer Science, Texas A&M University-Kingsville, TX 78363-8202, USA. ' Department of Computer Science, Auburn University, Auburn 36849-5347, USA. ' Department of Electrical Engineering and Computer Science, Texas A&M University-Kingsville, TX 78363-8202, USA

Abstract: In this paper, we introduce a novel dynamic cache management approach, i.e., a cache partitioning system (CaPaS), to optimise security levels in contemporary cluster storage systems for data-intensive applications with acceptable disk response times. CaPaS utilises an adaptive security control mechanism to adjust for varying workload conditions initiated by client-issued disk requests as well as security requirements and contains a cache partitioning scheme, a response-time estimator, and an adaptive security quality controller. The CaPaS algorithm solves a non-linear optimisation problem to increase the quality of security of disk requests in cluster storage systems while ensuring the disk requests to be completed within their desired response times. The efficiency of the proposed algorithm has been tested with the newly defined performance indicators on a cluster storage system including CaPaS, eight cryptographic and seven integrity services. Numerical simulation results demonstrate that the CaPaS algorithm improves the overall system performance enormously over two traditional baseline cache management algorithms.

Keywords: cache partitioning; security levels; security overheads; satisfied ratio; high performance architectures; cache replacement; cluster storage systems; dynamic cache management; simulation.

DOI: 10.1504/IJHPSA.2011.045506

International Journal of High Performance Systems Architecture, 2011 Vol.3 No.4, pp.216 - 232

Received: 24 Aug 2011
Accepted: 15 Nov 2011

Published online: 21 Mar 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article