Scheduling Microservice-Based Applications Across Distributed Edge Clusters

Cloud Computing and Edge Computing are becoming the standard for the development and deployment of software applications. Scheduling frameworks for resource management of an infrastructure nodes play a key role in supporting these approaches. This thesis describes \name, a centralized gang scheduler that can handle microservice applications taking into account both the resource and cost requirements of the single components and those derived by the communication between multiple components. Testing and validation demonstrated an average of 10 ms scheduling time, per application, on infrastructures composed by 100 nodes. Increasing the number of nodes to 1000 the scheduler kept the average scheduling time under 100 ms. To provide more value to the result a performance comparison with the state of the art Firmament scheduler was conducted. The measuring showed a performance improvement in scheduling time and cost management.