Authors: Randy Schauer; Anupam Joshi
Addresses: University of Maryland Baltimore County (UMBC), Baltimore, Maryland, USA ' University of Maryland Baltimore County (UMBC), Baltimore, Maryland, USA
Abstract: Large-scale distributed systems are playing an ever increasing role in computational research, modelling and simulation, information processing, and application hosting, via traditional high performance computing, cloud and data intensive systems. The continuous management of such systems is a critical consideration when focusing on reliability, availability, and security. As the size and complexity of these systems continue to grow, it becomes increasingly difficult to track the multitude of parameters required to ensure optimal performance from the system, especially in those systems that have been built through expansion over time as heterogeneous architectures and not as a single homogeneous platform. In this paper, we discuss the use of Markov logic networks, in a distributed multi-agent system to provide an effective means of managing these parameters. We showcase an example for analysis, providing a comparative illustration of how this approach is resolving differences between various system nodes after identifying potential configuration issues.
Keywords: autonomic computing; distributed systems; system configuration; configuration management; Markov logic networks; MLNs; probabilistic inference; high performance computing; multi-agent systems; MAS; agent-based systems; configuration parameters; system management.
International Journal of Autonomic Computing, 2016 Vol.2 No.2, pp.137 - 154
Received: 12 Aug 2015
Accepted: 03 Sep 2016
Published online: 31 Jan 2017 *