MULTI-STEP SERVICE OF REQUESTS CRITICAL TO QUEUEING DELAYS IN MULTI-TIER SYSTEMS
Read the full article
For citation: Bogatyrev V.A., Bogatyrev S.V. Multi-step service of requests critical to queueing delays in multi-tier systems. Scientific and Technical Journal of Information Technologies, Mechanics and Optics, 2017, vol. 17, no. 5, pp. 872–878 (in Russian). doi: 10.17586/2226-1494-2017-17-5-872-878
Subject of Research.The paper deals with effectiveness analysis of redundant service of requests time-critical to total gradual waiting in the queues of multi-tier clustering nodes. We search for design solutions on the organization of multi-stage redundant service of copies of requests in multi-tier cluster that makes it possible to increase the probability of timely servicing of requests critical to the total waiting time for all system levels (stages of maintenance). Method. We modeled the multi-stage redundant service of copies of requests in multi-tier cluster. The redundancy effectiveness is estimated by the probability of not exceeding the total phased waiting time in queues of nodes for all levels specified for the maximum allowable time. We selected the best options for the organization of redundant requests; the request is considered successful if at least for one accurately executed copy, total waiting time of the stages in the queues of the nodes at all levels of the cluster does not exceed the maximum allowable waiting value. Main Results. The experiments based on calculations carried out with the use of the proposed models on the three-tier cluster example gave the possibility to elucidate the dependence of probability of not exceeding the allowable total request time on the intensity of requests, the multiplicity of redundancy and the maximum allowable cumulative waiting time of requests in the queues of the nodes at all cluster levels. We have shown the existence of efficient redundant service of requests in multi-tier clusters, and the optimal multiplicity of redundancy of request copies depending on the system load and constraints on the allowable total time of staged waiting in the queues of the nodes in the clusters at all levels. Practical Relevance. The proposed models can be used to assess the reliability and timeliness of the service processes that are critical to request waiting time, as well as in the justification of selection of disciplines and service parameters in multi-tier computer systems, including real-time ones.
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License