Peculiarities of Modeling a Specialized Computing System

When developing distributed computing systems with parallel data processing, there is a problem of assessing the impact of workload values and structure on its performance indexes. One of the key points in this problem is to assess the impact of various prioritization disciplines on the time characteristics of emerging request queues in the system, for which statistical methods of data analysis are used.

The goal of this study is to develop a method for constructing a simulation model that will allow estimating the time characteristics of the system depending on changing workload values and a priorityprocessing algorithm. The method is based on the joint use of the developed simulation model, which describes in detail the functioning of the system of the considered class in time, taking into account conflict situations arising during parallel processing of information, and experimentally obtained individual time characteristics of the system.

Materials and methods. The model is implemented in the GPSS language. All stages of applying the presented method are considered. Examples of workload for modeling are given. Justifications for using the presented data, as well as the principles by which they were selected, are given. For the analyzed class of problems, simulation modeling of the computing system was carried out. During the construction of the simulation model of the system a specialized data acquisition device as source of requests; a switch for which a request queue with different priorities is simulated; a data processing device, which is the final recipient of the data were selected as simulation functional nodes.

The types of algorithms used to solve the request prioritization problem are taken from common prioritization algorithms typical for the Quality of Service (QoS), used in modern switching equipment. Three prioritization algorithms were considered: without using priorities as a standard; priority queue; Weighted Round Robin as a more complex algorithm.

Data on the processing time of various types of requests were obtained experimentally using the Wireshark network traffic analysis tool. The obtained times, as well as the intensity of requests for request processing and the ratio of requests of different types are parameters of the created model and can be changed to simulate another system with a similar architecture.

Results. Based on the analysis of the obtained modeling results, the influence of various disciplines for processing request priorities in queues on the system performance indexes is shown. Regenerative model analysis method is used to analyze the obtained data. The obtained method allows for a detailed analysis of the system’s time characteristics, taking into account the prioritization of requests when they are processed in queues.

Conclusion. The conducted research analysis shows the impossibility of obtaining these metrics by means of analytical modeling, which emphasizes the novelty of the study. The method obtained during the study is used in the development of systems of the presented class, which emphasizes its practical significance and relevance. 

1. Pavskiy V.A., Pavskiy K.V. Mathematical model for calculating reliability indicators of scalable computing systems taking into account switching time. Izvestiya YUFU. Tekhnicheskiye nauki = Bulletin of SFedU. Technical sciences. 2020; 2(212): 134–145. DOI: 10.18522/2311-3103-2020-2-134-145. (In Russ.)

2. Brekhov O.M., Moraru V.A. Analytical modeling of a data flow-controlled computer. Avtomatika i telemekhanika = Automation and Telemechanics. 1993; 12: 165–178. (In Russ.)

3. Afanas’yev A.P., Posypkin M.A., Khritankov A.S. Analytical model for evaluating the performance of distributed systems. Programmnyye produkty i sistemy = Software products and systems. 2009; 4: 60-64. (In Russ.)

4. Leont’yev A.S., Zhmatov D.V. Study of probabilistic and temporal characteristics of wireless networks with the CSMA/CA access method. Russian Technological Journal = Russian Technological Journal. 2024; 12(2): 67−76. DOI: 10.32362/2500-316X-2024-12-2-67-76. (In Russ.)

5. Leont’yev A.S. Development of analytical methods, models and techniques for analyzing local area networks. Teoreticheskiye voprosy programmnogo obespecheniya: Mezhvuzovskiy sbornik nauchnykh trudov = Theoretical issues of software: Interuniversity collection of scientific papers. Moscow: MIREA; 2001: 70–94. (In Russ.)

6. Lovina A.P. Review of the simulation modeling method [Internet]. Sovremennaya tekhnika i tekhnologii = Modern technology and technology. 2017: 5. Available from: https://technology.snauka.ru/2017/05/13530 (cited 17.12.2021). (In Russ.)

7. Brekhov O.M., Zvonareva G.A., Ryabov V.V. Features of the development and analysis of a simulation model of a multiprocessor computing system. Otkrytoye obrazovaniye = Open education. 2017; 3: 48-56. (In Russ.)

8. Kreyn M., Lemuan O. Vvedeniye v regenerativnyy metod analiza modeley = Introduction to the regenerative method of model analysis. Moscow: Science; 1982, 104 p. (In Russ.)

9. Zvonarova G.A., Buzunov D.S. Using simulation modeling to assess the temporal characteristics of a distributed computing system. Otkrytoye obrazovaniye = Open education. 2022; 26(5): 32-39. DOI: 10.21686/1818-4243-2022-5-32-39. (In Russ.)

10. Khoring CH. A Standard for the Transmission of IP Datagrams over Ethernet Networks = Standard for the Transmission of IP Datagrams over Ethernet Networks [Internet]. Available from: https://datatracker.ietf.org/doc/html/rfc894. (In Russ.)

11. Nichols K, Bleyk S. Definition of the Differentiated Services Field (DS Field) in the IPv4 and IPv6 Headers [Internet]. 1998. Available from: https://datatracker.ietf.org/doc/html/rfc2474.

12. Dordal L.P. An Introduction to Computer Networks. Chicago, Loyola University. 2020. [Internet]. Available from: https://intronetworks.cs.luc.edu/current/html/index.html.

13. Buzunov D.S., Zvonareva G.A. Modeling of a specialized computing system with parallel information processing. 20-ya Mezhdunarodnaya konferentsiya «Aviatsiya i kosmonavtika» =20th International Conference «Aviation and Cosmonautics». 2021:183-184. (In Russ.)

14. Davies J. The Cable Guy TCP Receive Window Auto-Tuning, TechNet Magazine; 2007.

15. Bekhterev A. Kachestvo setey peredachi dannykh. Programmnyye i apparatnyye izmereniya = Quality of data transmission networks. Software and hardware measurements [Internet]. Available from: https://habr.com/ru/articles/250821/. (In Russ.)

16. Brekhov O.M., Zvonareva G.A., Korneyenkova A.V. Imitatsionnoye modelirovaniye = Simulation modeling. Moscow: MAI; 2015. 324 p.

17. Shrayber T. Modelirovaniye na GPSS = Modeling on GPSS. 1980. 592 p. (In Russ.)

18. Kling T., The impact of transport protocol, packet size, and connection type on the round trip time. Blekinge Institute of Technology. 2017.

19. Siemon D., Queueing in the Linux netwok stack. Linux Journal: Networking. 2013.

20. Ross K., Kurose J. «Delay and Loss in Packet-Switched Networks». 2012.