Continuous monitoring methods to achieve resiliency for virtual machines

This dissertation describes monitoring methods to achieve both security and reliability in virtualized computer systems. Our key contribution is showing how we can perform continuous monitoring and leverage information across different layers of a virtualized computer system to detect malicious attacks and accidental failures. For monitoring software running inside a virtual ma- chine, we introduce HyperTap and Hprobes, which are out-of-VM monitoring frameworks that facilitate detection of security and reliability incidents oc- curring inside a VM. For monitoring the hypervisor, we introduce hShield, a Control-Flow Integrity (CFI) enforcement method to detect VM-escape at- tacks. HyperTap, Hprobes, and hShield create a complete chain-of-trust for the entire virtualization software stack