Optimal computational resource allocation for implementation of central bank digital currency services

Abstract

The implementation of the central bank digital currency (CBDC) as a distributed computing service with interconnected components is considered under conditions of high load. When the overall performance of the service depends on the maximum throughput of one of its components the necessity to redistribute resources between the service components arises. The deviation of the sum of CPU resource utilization of each service component from the given resource volume dependent on given workload is introduced as an objective function for finding the optimal solution. The optimization goal is to find the allocation of computational resources between the involved service components while minimizing the objective function. A method for optimal resource allocation has been developed, which is based on the described optimization problem and identification of functional relation of CPU resource utilization of each service component to the given workload, that is achieved with load testing using scalable experimental setup consisting of six CBDC service components. The specifics of data processing during operations with CBDC were considered in the process of the models’ construction. The results obtained demonstrate the effectiveness of the developed models and the method for finding the optimal resource allocation.