uServ: compartilhando recursos computacionais ociosos para execução de serviços Cloud Native na borda da Internet

Abstract

This thesis explores the feasibility of utilizing idle computational resources from personal computers (PCs) to execute cloud-native services by proposing the uServ system. uServ aims to monetize these idle resources, allowing PC owners to leverage their underutilized devices while microservice developers deploy their applications closer to end-users. The central problem addressed is the difficulty of effectively connecting the demand for and supply of computational resources capable of executing cloud-native services, and the thesis proposes innovative solutions to overcome this challenge. uServ operates through a distributed architecture that efficiently manages and allocates resources. It uses specific algorithms to identify idle resources, homogenize and price these resources, establish dynamic regions, and distribute microservices within these regions. In doing so, the system creates regional markets controlled by auction processes that aim to define a fair price for the final allocation of resources. Simulation results confirmed that uServ is a promising solution for executing microservices at the network edge, offering economic benefits to both providers and consumers. Providers can earn financial returns from their idle PCs without additional effort, while consumers can reduce latency and costs of their services by leveraging uServ’s decentralized infrastructure. The literature review highlighted the research gap that uServ fills by integrating cloud-native services with the idle infrastructure of personal computers. The practical applicability of the system was demonstrated, showing it to be adaptable to various strategic needs and providing a flexible and economically advantageous solution.