DOI 10.17586/0021-3454-2023-66-8-646-651
UDC 004-75
DETERMINING THE CLUSTER SIZE AND THE NUMBER OF REPLICAS OF HIGHLY LOADED INFORMATION SYSTEMS
Saint Petersburg Electrotechnical University “LETI”, Saint Petersburg, 197376, Russian Federation; Russian State Hydrometeorological University, Saint Petersburg, 195196, Russian Federation; Head of Chair
E. D. Arkhiptsev
Saint Petersburg Electrotechnical University “LETI”, Saint Petersburg, 197022, Russian Federation; Post-Graduate Student
N. S. Karmanovskiy
ITMO University, Saint Petersburg, 197101, Russian Federation; Associate professor
Read the full article
Reference for citation: Tatarnikova Т. М., Arkhiptsev E. D., Karmanovskiy N. S. Determining the cluster size and the number of replicas of highly loaded information systems. Journal of Instrument Engineering. 2023. Vol. 66, N 8. P. 646—651 (in Russian). DOI: 10.17586/0021-3454-2023-66-8-646-651.
Abstract. The problem of ensuring the quality of customer service for highly loaded information systems is discussed. Existing solutions for scaling highly loaded information systems are considered. A procedure for determining the required number of nodes in a computing cluster based on the queuing theory is proposed. Results of a computational experiment on choosing the size of a computational cluster to ensure the average time for a user request executing to be below the established allowable time, are presented. A model for determining the number of replicas is proposed, in which, at a given intensity of incoming update requests, it is possible to avoid the critical level of the write-write conflict.
Abstract. The problem of ensuring the quality of customer service for highly loaded information systems is discussed. Existing solutions for scaling highly loaded information systems are considered. A procedure for determining the required number of nodes in a computing cluster based on the queuing theory is proposed. Results of a computational experiment on choosing the size of a computational cluster to ensure the average time for a user request executing to be below the established allowable time, are presented. A model for determining the number of replicas is proposed, in which, at a given intensity of incoming update requests, it is possible to avoid the critical level of the write-write conflict.
Keywords: high-load information system, customer service quality, scaling, replication, mathematical modeling
References:
References:
- Tuzov A.V. Bulletin of The South Ural State University. Series: Computer Technologies, Automatic Control, Radio Electronics, 2018, no. 3(18), pp. 5–11. (in Russ.)
- Bogatyrev V.A., Derkach A.N., Aleksankov S.M. Studies in Systems, Decision and Control, 2020, vol. 260, рр. 11–21.
- Аmirov S.N. International Journal of Open Information Technologies, 2020, no. 8(8), pp. 32–47. (in Russ.)
- Gorodnichev M.G., Kochupalov A.E. The Eurasian Scientific Journal, 2018, no. 4(10), pp. 1–32. (in Russ.)
- Podolny V.P. Arkhitektura vysokonagruzhennykh sistem. Sistemy sbora informatsii, raspredelennyye sistemy upravleniya, sistemy real'nogo vremeni (Architecture of Highly Loaded Systems. Information Collection Systems, Distributed Control Systems, Real-Time Systems), Moscow, 2022, 160 р. (in Russ.)
- Cvyashhenko E.V. Information Systems and Technologies, 2016, no. 1, pp. 74–83. (in Russ.)
- Bogatyrev V.A., Lisichkin D.E. Software & Systems, 2019, no. 2, pp. 214–220. (in Russ.)
- Sovetov B.Ya., Tatarnikova T.M., Poymanova E.D. Information and Control Systems, 2020, no. 5(108), pp. 43–49, DOI: 10.31799/1684-8853-2020-5-43-49.
- Tatarnikova T.M., Volskiy A.V. Information and Control Systems, 2018, no. 3(94), pp. 54–60, DOI: 10.15217/issn1684-8853.2018.3.54. (in Russ.)
- Tatarnikova T.M., Elizarov M.A. Journal of Instrument Engineering, 2017, no. 2(60), pp. 150–157, DOI: 10.17586/0021-3454-2017-60-2-150-157. (in Russ.)
