ANALYTICAL MODEL OF THE IMPACT OF PARALLELISM ON THE PERFORMANCE OF AN ORDERED EVENT DELIVERY SYSTEM WITH AT-LEAST-ONCE SEMANTICS

DOI: 10.31673/2409-7292.2026.011987

Abstract

In modern distributed information systems based on microservice and event-driven architectures, asynchronous
mechanisms for exchanging messages through event brokers are widely used. One of the important problems of such systems
is ensuring ordered event processing when using at-least-once delivery semantics, which guarantees message delivery at least
once, but allows for its repeated arrival. Under the conditions of parallel event processing, additional factors arise that affect
system performance, in particular the costs of idempotence checking, message reprocessing, event buffering to restore their
correct order, the use of backup delivery channels, and competition for shared resources. Traditional models for analyzing the
performance of parallel systems, such as the ideal scaling model and the Amdahl model, do not take into account the specific
features of event-driven message delivery systems. The aim of the work is to develop an analytical model of the impact of
parallelism on the performance of an ordered event delivery system with at-least-once semantics. The paper proposes a
generalized mathematical model that takes into account the costs of idempotent processing, message re-delivery, event buffering
delays, the use of backup transmission channels, and competition for shared system resources. In addition, the model takes into
account the uneven distribution of the load between event keys, which allows us to estimate the effective level of parallelism
and determine the limits of system scalability. Based on the proposed model, an analytical dependence of the system throughput
on the number of parallel processors was obtained and the efficiency coefficient of parallel resource use was determined. A
comparative analysis was conducted with classical scaling models, which showed a more accurate description of the behavior
of event-driven systems. The practical value of the paper lies in the possibility of using the model to predict the performance
and optimize the configuration parameters of distributed microservice systems.
Keywords: event-driven architecture, parallel event processing, system throughput, microservice systems, at-least-once
delivery, distributed system performance, information technologies.

References
1. McGlohon N., Carothers C. Toward Unbiased Deterministic Total Orderings of Parallel Simulations with
Simultaneous Events. Proceedings of the Winter Simulation Conference, 2021. DOI: 10.1109/WSC52266.2021.9715459.
2. Bachan J., Ye J. C., Jiang X. et al. Devastator: A Scalable Parallel Discrete Event Simulation Framework for
Modern C++. Proceedings of the ACM, 2024. DOI: 10.1145/3615979.3656061.
3. Bhupatiraju R. V. Event-Driven Architecture for Payment Failover and Redundancy: A Framework for HighAvailability Financial Transaction Processing. European Modern Studies Journal, 2025. DOI: 10.59573/emsj.9(5). 2025.
75.
4. Poudel P., Rai S. K., Guragain S. Ordered Scheduling in Control-flow Distributed Transactional Memory.
Theoretical Computer Science, 2024. DOI: 10.1016/j.tcs.2024.114463.
5. Andelfinger P., Köster T., Uhrmacher A. M. The Window Racer Algorithm for Parallel Discrete Event
Simulation. Proceedings of the ACM, 2023. DOI: 10.1145/3573900.3591115.
6. Poudel P., Rai S. K., Guragain S., Sharma G. Ordered Scheduling in Control-Flow Distributed Transactional
Memory. Lecture Notes in Computer Science, Springer, 2023. DOI: 10.1007/978-3-031-24848-1_5.
7. Batista E., Coelho P., Alchieri E. et al. FlexCast: An Efficient Atomic Multicast Protocol for Distributed
Systems. Proceedings of the ACM, 2023. DOI: 10.1145/3590140.3629122.
8. Poudel P., Rai S., Sharma G. Processing Distributed Transactions in a Predefined Order. Proceedings of the
ACM, 2021. DOI: 10.1145/3427796.3427819.
9. Eker A., Arafa Y., Badawy A. A. et al. Load-Aware Dynamic Time Synchronization in Parallel Discrete Event
Simulation. Proceedings of the ACM, 2021. DOI: 10.1145/3437959.3459249.
10. McGlohon N., Carothers C. Deterministic Event Ordering Techniques for Parallel Discrete Event Simulation
Systems. ACM Digital Library.
11. V. Zhebka, et al., Methods for predicting failures in a smart home, in: Digital Economy Concepts and
Technologies Workshop, vol. 3665, 2024, 70–78.