Resumen
The subject matter of the article is development of the models and methods of load and resource balancing and reallocation in cloud computing systems based on the infrastructure as a service model. The goal of the work is to increase the efficiency of available resources usage in cloud computing systems (such as RAM, disk space, CPU, network) by developing the model for adaptive management of resource reallocation. This will allow new virtual machines to be launched with minimal performance degradation for already running applications. The following tasks were solved in the article: development of an complex approach to manage resource reallocation in cloud systems, including decomposition of the cloud computing system into zones (based on the defining features of the resources provided in each zone), initial resource allocation (based on the hierarchy analysis method) and resources reallocation within cloud computing system (based on the developed method); development of a method for computing resources reallocation in cloud computing systems; evaluation of the effectiveness of the developed method. To solve the set tasks, approaches and methods of dynamic load balancing were used, as well as methods of theoretical research, which are based on the scientific provisions of the theory of artificial intelligence, static, functional and system analysis. The following results were obtained ? on the basis of existing load balancing methods in cloud computing systems analysis, the main features of existing resource allocation methods were identified, their advantages and disadvantages were given. On the basis of the conducted analytical study, the necessity of improving the existing methods of resource reallocation has been proved. A method and an algorithm for computing resources reallocation within cloud computing systems have been developed. This make it possible to reduce the values of the coefficient of computing resources uneven usage while minimizing the cost of moving them. The results obtained have been confirmed by experiments carried out using software for creating private infrastructure cloud services and cloud storages. Conclusions: the improvement of the reallocation and load balancing method in cloud computing systems has increased the ability of these systems to launch new virtual machines with a minimum decrease in the performance of already running applications.