Дослідження доступності у хмарних сервісних системах

Abstract

Наведено результати числового моделювання сервісної доступності хмарних сервісних систем для різних їх конфігурацій, характеристик трафіку та режимів роботи, а також проведено оцінювання ефективності застосування запропонованого методу підвищення показників доступності вузлів хмарної платформи. This paper is devoted to the numerical investigation of the availability at cloud service systems. In this paper criteria and constraints calculations were performed and obtained results were analyzed for synthesis purposes of distributed service platforms based on the cloud service-oriented architecture such as availability and system performance index variations by defined set of the main parameters. The method of synthesis has been numerically generalized considering the type of service workload in statistical form by Hurst parameter application for each integrated service that requires implementation within the service delivery platform, which is synthesized by structural matching of virtual machines using combination of elementary servicing components up to functionality into a best-of-breed solution. As a result of restrictions from Amdahl’s Law the necessity of cloud-networks clustering was shown, which makes it possible to break the complex dynamic network into separate segments that simplifies access to the resources of virtual machines and, in general, to the "clouds" and respectively simplifies complex topological structure, enhancing the overall system performance. In overall, proposed approaches and obtained results numerically justifying and algorithmically describing the process of structural and functional synthesis of efficient distributed service platforms, which under process of their configuring and exploitation provides an opportunity to act on the dynamic environment in terms of comprehensive services range and nomadic users’ workload pulsing. It was also proved that using proposed approach and algorithm promotes us to determine structural-topological changes operatively in dynamics. It allows improving workload balancing in the servicing nodes, avoiding the proximity of workflows and components of complex services to structural irregularities. The whole spatial workflow distribution on the servicing network is optimized and smoothed in accordance with operatively determined workload and structural features. Cloud system service node’s availability is improved. Prospective usage of proposed algorithm is looking effectively within distance-vector routing protocols (such as BGP) and, in particular at hyperbolic virtual coordinates mapping, where hyper-aggregation of the flows on the edges of Poincare disk’s irregularities is observed.