Improving of Totally Ordered Multicast Algorithm for Distributed Architecture Data Processing Systems

: pp. 241 - 247
Lviv Polytechnic National University
Lviv Polytechnic National University
Lviv Polytechnic National University

The gradual shift from individual to shared service delivery platforms brings distributed service delivery systems to the forefront and the problems associated with their proper functioning. Performance of totally ordered multicasting method was modeled and studied in this paper. The analysis revealed that current realization of the method is not suitable for processing non-FIFO-queue events, thus is not able to function adequately in modern high- intensity flow of service requests. Accordingly, improved method is proposed. The block diagram of this method is depicted in the paper. The simulation process is specified. Performance of modified totally ordered multicasting method was modeled according to this process and it is shown that proposed changes allow decreasing of failure rate in modern distributed service delivery systems. The paper shows the use of totally ordered multicasting method in real systems is complicated by the fact that it only works well in the channel FIFO. Because of this disadvantage, we propose improved method of totally ordered multicasting, which enables the synchronization process to survive, even if the message of sender process is lost. The improvement is to limit the waiting time for a response from all the processes addressed. This is to reduce the probability of failure of the process. However, the signaling information amount increases that circulates in the data network.

1. Tanenbaum, E. and van Steen, M. (2003), Distributed Systems. Principles and Paradigms, Piter. 2. Toporkov, V. (2004), Distributed computing models, Fizmalit. 3. Fidge, C. (1991), “Logical time in distributed computing systems”, IEEE Computer, Vol. 24, No. 8, pp. 28–33. 4. Fowler, J. and Zwaenepoel, W. (1990), “Causal distributed breakpoints”, Proceedings of the 10th International Conference on Distributed Computing Systems, pp.134—141. 5. Lamport, L. (1978), “Time, clocks and the ordering of events in a distributed system”, Communications of the ACM, no. 21, pp. 558–564. 6. Rauber, T. and Rünger, G. (2013), Parallel Programming For Multicore and Cluster Systems, Springer. 7. Singhal, M. and Kshemkalyani, A. (1992), “An efficient implementation of vector clocks”, Information Processing Letters, no. 43, pp. 47–52.