Models for Cloud System Availability Assessment Considering Attacks on CDN and ML Based Parametrization

Abstract
The article proposes a method for assessing the availability of the cloud system taking into account the variable dynamics of attacks on vulnerabilities Content Delivery Network (CDN), The architecture of the cloud system for video hosting services is detailed, on the basis of which an example of simulation in the conditions of cyberattacks, software and hardware failures is given. An availability model based on the Reliability Block Diagram (RBD), a Markov model (MMC) with constant parameters of failure and recovery rates, and a multifragment (MFM) model with a variable parameter that estimates the probability of attacks have been developed and studied. Two scenarios of events that affect the availability of the system are considered: the first - in the absence of attacks on the CDN component; the second - in attacks that cause an increase in the CDN failure rate to the limit level. A comparative analysis of RBD, MMC and MFM and assessment of discrepancies in the simulation results were performed. The use of Big Data analytics and ML tools is proposed for parametrization of models. The obtained simulation results can be used not only by users of cloud systems, but also by Cloud Service Providers (CSP) to improve planning procedures and risk assessment of failures.
Description
Ponochovnyi Y., Ivanchenko O., Kharchenko V., Udovyk I. Baiev E. Models for Cloud System Availability Assessment Considering Attacks on CDN and ML Based Parametrization // Computational Linguistics and Intelligent Systems. Proceedings of the 6th International Conference on COLINS 2022. Volume I: Workshop. Gliwice, Poland, May 12-13, 2022, / V. Lytvyn et al (edits), CEUR Workshop Proceedings, Volume 3171, 2022, pp. 1149-1159.
Keywords
Availability assessment, multifragment markoy models, cloud system, reliability block diagrams, attack on content delivery network, machine learning for model parametrization
Citation