Explosive growth in shared-data-center computing has increased the need for fair resource-allocation policies and mechanisms that provide quality-of-service (QoS) guarantees for tenants sharing a physical infrastructure. The focus today is on virtualizing entire applications with integrated compute, storage, and networking resources. This requires automated techniques for dynamically slicing and aggregating the underlying physical infrastructure, to create ensembles of virtual resources with flexible mixes of different resource types.
The project involves the development of policies, algorithms, and scheduling mechanisms to optimize the joint allocation of multiple heterogeneous resources. Models for fair sharing of multiple resources suitable for these systems will be defined, and new resource allocation algorithms that maximize consolidation ratios will be developed and evaluated. The results will advance the development of the fast-growing cloud IT infrastructure by increasing resource efficiency for the service providers, and providing better QoS guarantees and isolation for the tenants. The project will also result in the education and training of students in a topical and important sector of Information Technology.