Development of an algorithm for resource allocation in distributed systems based on two- criteria process assessment

One of the most important factors that are affecting of quality of processes service in a distributed system is the way of managing shared resources. At the same time, it is necessary to take into account that each process (task) has a set of characteristics, the complex accounting of which makes it possible to increase the efficiency of executing processes. Among the most important characteristics are the execution time of the process and its significance for solving system-wide tasks. This article is devoted to the development of a resource allocation algorithm based on a two-criteria process assessment. The priority and order of tasks execution is determined based on the importance weights formed by the PROMETHEE II multicriteria decision-making method. The paper describes the features of the application of this method to solve the problem and design an algorithm for the resource’s allocation based on a two-criteria assessment of processes. The algorithm provides for the possibility of interrupting the service of processes and forming a queue based on the importance weights. To automate the resource planning process, a software product has been developed that implements the stages of the algorithm. The calculations have shown that the proposed algorithm improves the quality of management of distributed systems, making the resource planning process more flexible and efficient. The approach described in the work is universal and can be extended for the case of an arbitrary number of criteria for evaluating processes.

1. Thamsen L., Verbitskiy I., Beilharz J., Renner T., Polze A., Kao O. Dynamically Scaling Distributed Dataflows to Meet Runtime Targets. Proc. Int/ Conf. Cloud Comput Technol Sci CloudCom. 2017;37:146-153.

2. Silberschatz A., Galvin P.B., Gagne G. Operating Systems Concepts. New York: John_Wiley_&_Sons. 2008.

3. Thamsen L., Renner T., Kao O. Continuously Improving the Resource Utilization of Iterative Parallel Dataflows. Proceedings of the 6th International Workshop on Big Data and Cloud Performance, ser. DCPerf 2016. IEEE. 2016;1(4):1-6.

4. Castillo G., Rouskas N., Harfoush K. Efficient QoS resource management for heterogeneous Grids. 22nd. IEEE International Parallel and Distributed Processing Symposium (IPDPS'08), Miami, Florida, US. 2008:1-15.

5. Jiang H., Ni T. PB-FCFS-a task scheduling algorithm based on FCFS and backfilling strategy for grid computing. Proceedings of Joint Conferences on Pervasive Computing (JCPC). 2009:507- 510.

6. Triantaphyllou E. Multi-Criteria Decision Making Methods in Multi-Criteria Decision Making Methods, a Comparative Study. Applied Optimization. 2000; 44(1):5-21.

7. Chen L., Xu Z., Wang H., Liu S. An ordered clustering algorithm based on K-means and the PROMETHEE method. International Journal of Machine Learning and Cybernetics. 2018; 9(6):917-926.

8. Chakraborty S., Yeh C. H. A Simulation Based Comparative Study of Normalization Procedures in Maldistributed Decision Making. Int. Conf. on Artificial Intelligence, Knowledge Engineering and Data Bases.2007:102-109.

9. Mareschal B., Brans J.-P. PROMETHEE Methods. International Series in Operations Research and Management Science.2014;78(2):163-195.

10. Yu L., Chen L., Cai Z., Shen H., Liang Y., Pan Y. Stochastic Load Balancing for Virtual Resource Management in Datacenters. IEEE Transactions on Cloud Computing. 2018; 8(2):459-472.