For example,Бобцов

ORGANIZATION OF SIMULATION EXPERIMENTS IN THE DESIGN OF DISTRIBUTED COMPUTER SYSTEMS WITH REDUNDANT DATA TRANSMISSION

Annotation

Subject of Research.We study the ways of multipath distribution of copies of requests (packets) through the network and their redundant servicing by servers of a given cluster. The alternatives under consideration differ in the way of path selection for a packet in the network having routes with joint switching nodes (static or dynamic) and in the condition under which those copies exit early from the system as a result of their losses in the communication channels, the emergence of bit errors, the overflow of transmitting and receiving buffers of the network and computing equipment, or the combination of these reasons. Method. We consider the efficiency of redundant transmissions, defined as the probability of faultless and timely delivery of at least one redundant copy to the cluster and dependent on the intensity of the flow of incoming requests. Based on the alternatives described, a set of simulation models with parameters representing the requirements for the system under simulation is created with AnyLogic 7 simulation environment. The results of simulation experiments are being written to the Microsoft SQL Server 2008 database, undergo transformations, and then are exported into the application program developed in this study. Main Results. Organization of simulation experiments involving accumulating and processing of their results with the tools of the chosen database management system is proposed. After being handled, those results are used to form the basis for analysis of the obtained dependencies with third-party software applications, such as Microsoft Excel or the one implemented in this work, to determine the scope of efficiency of redundant transmissions and to automate the search for the optimal order of redundancy. The simulation results that confirm the efficiency of multipath distribution of copies of requests are presented. Practical Relevance.The proposed models and tools enable one to justify the choice and optimize project solutions to the problem of improving functional reliability of distributed computer systems with redundant transmission and servicing of requests.

Keywords

Постоянный URL

Articles in current issue

CONTENT © ITMO UNIVERSITY 1993-2024