Russian Federation
The article presents the development and analysis of a stochastic model of cascade latency in distributed systems of 2026 based on nonlinear diffusion equation with multiplicative feedback. The authors classify cascade mechanisms, formalize amplification factor α >1 and tail latency via super-heavy-tailed law, implementing vectorized simulation of millions of requests with importance sampling and online prediction via phase diagrams. The uniqueness of the work lies in predicting avalanche slowdowns seconds before critical phase, preventing >80% cascades with automatic stabilizers, reducing p99.9 tail latency by orders of magnitude and increasing throughput by an order.
cascade latency, tail latency, nonlinear diffusion, microservice chains, predictive throttling
1. Mochalov, V. P. Algoritm dinamicheskogo raspredeleniya i balansirovki nagruzki v raspredelennyh oblachnyh vychisleniyah / V. P. Mochalov, N. Yu. Bratchenko, D. V. Gosteva // Modelirovanie sistem i processov. – 2024. – T. 17, № 1. – S. 92-102. – DOIhttps://doi.org/10.12737/2219-0767-2024-17-1-92-102. – EDN EWMPYM.
2. Haritonov, A. V. Priblizhennaya ocenka zaderzhek v komp'yuternoy sisteme s konteynernoy virtualizaciey // Informatika i avtomatika. – 2025. – T. 24, № 3. – S. 917-925. – URL: https://www.mathnet.ru/links/3bb517bfcc71c951793066d9968cc9cf/trspy1377.pdf (data obrascheniya: 18.02.2026).
3. Grigor'ev, V. P. Analiz struktury zaderzhek peredachi informacii v kommunikacionnoy srede klasterov // Russkie superkomp'yuternye dni. – 2022. – S. 1-10. – URL: https://russianscdays.org/files/pdf/546.pdf (data obrascheniya: 18.02.2026).
