Optimizing the architecture of distributed shared memory systems in real time applicatons

ABSTRACT

The increasing demands of distributed and parallel applications require sufficiently high-performance distributed shared memory algorithms capable of providing the service almost steadily. Past and recent decades witnessed the introduction of number of algorithms and strategies aim at tuning and optimizing the performance of distributed systems and at extending the scalability of the systems. This paper presents a new strategy that adjoins an intermediate level of control to the system to scale up the performance via relieving the overhead at the main server. This strategy can be implemented at both setup- and run-time. The findings indicate that the performance of DSM systems can be improved significantly with this strategy.

REFERENCES

• 1: C. Amza, A. L. Cox, S. Dwarkadas, J. Li-Jie, K. Rajamani, W. Zwaenepoel, "Adaptive protocols for software distributed shared memory", Proceedings of the IEEE, vol.87, no.3, March 1999, pp. 467--75. Publisher: IEEE, USA.
• 2: F. Mueller, "Distributed Shared-Memory Threads: DSM-Threads", Workshop on Run-Time systems for Parallel Programming, Apr 1997.
• 3: Khalil M., Peytchev E., "Traffic Telematic Computing Framework based on Non-Locking and Housekeeping Distributed Shared Memory Algorithm", Sixth United Kingdom Simulation Society Conference (UKSIM2003), Apr. 2003, Emmanuel college, Cambridge, UK.
• 4: Helen D. Karatza , Ralph C. Hilzer, Epoch Load Sharing in a Network of Workstations, Proceedings of the 34th Annual Simulation Symposium (SS01), p.36, April 22-26, 2001
• 5: Helen D. Karatza , Ralph C. Hilzer, Parallel and distributed systems: load sharing in heterogeneous distributed systems, Proceedings of the 34th conference on Winter simulation: exploring new frontiers, December 08-11, 2002, San Diego, California

INDEX TERMS

Keywords:
distributed computing, heuristic, intermediate server, optimization, round-trip time