Tucano: A Service Scheduler and Load Balancer in a Distributed System

Abstract

As applications grow in complexity, their workloads grow heavier and more resource demanding. To address this challenge, a system of network-connected servers can be employed to distribute the processes from such applications, reducing performance loss caused by unequal resource sharing. This work introduces Tucano: a scheduler that leverages operating systems’ techniques to manage workloads in a distributed, container-oriented context. Additionally, Tucano implements a load balancer, that ensures the correct mapping of requests to their corresponding host servers. Through physical distribution and containerization, Tucano aims to explore an efficient and scalable strategy for managing multiple workloads.