Sharing-Aware Resource Management Algorithms for Virtual Computing Environments

Abstract

Virtualization technologies are ubiquitous in data centers around the world where providers consider operational costs and fast delivery guarantees for a variety of profitable services. Providers should consistently invoke measures for increasing the efficiency of their virtualized services in a competitive environment where fast entry to market, technology advancement, and service price differentials separate sustaining providers from antiquated ones. Therefore, providers seeking further efficiencies and profit opportunties should consider how their resources are managed in virtual computing environments which leverage memory reclamation techniques, specifically pagesharing; motivating the design of new sharing-aware resource management algorithms. In this Ph.D. dissertation, we design families of offline and online sharing-aware algorithms for resource management in virtual computing environments, and investigate their properties and their relationships to various sharing models. Our contribution innovates the design of new approximation algorithms offering relevant performance guarantees.