Towards Trustworthy Virtualisation: Improving the Trusted Virtual Infrastructure
Carl Gebhardt · 2011
Modern commodity platforms have become easy targets, which are increasingly plagued by malware exploiting legacy design weaknesses. Malware often abuses the large and feature-rich computing base, which forms the basis of modern commodity systems and inherently has to be trusted. In recent years, research has suggested employing machine virtualisation technology to provide isolation where commodity systems fail to do so. On one hand, hardware machine virtualisation support on commodity systems is a very recent technology, and with its novel technology allows for new creative security solutions. On the other hand, machine virtualisation changes many previous security assumptions about a platform and therefore creates new challenges itself. This thesis investigates machine virtualisation and trusted computing technology and outlines how those technologies could be utilised to move towards a more trustworthy virtualisation infrastructure. To achieve this, the thesis has been divided into three main parts. In the first part of this thesis, we describe how the hypervisor’s Trusted Computing Base could be reduced and, with new hardware advances, could be further strengthened. To achieve this, we reassess the definition of the Trusted Computing Base and illustrate how segregation of different code blocks could be enforced by hardware protection mechanisms. In the second part, we propose a novel scheme to protect the integrity and confidentiality of storage in a virtualised infrastructure. We discuss the implementation of a prototype for a secure, flexible and transparent virtual disk image. We base our concepts on trusted computing, utilising the Trusted Platform Module to efficiently deliver integrity assurance to virtual disk images, as well as enabling the owner to retain control over the disk image throughout its life-cycle.