Analysis of virtualization through co-processor

Abstract

Virtualization technology was created as a reliable time-sharing option in the early years. It is the most effective strategy to minimize IT costs while enhancing efficiency, speed, flexibility, and scalability in today's world. It is the process of producing a virtual (rather than real or physical) version of something in the field of computing, which encompasses virtual computer hardware platforms, computer network resources and storage devices. We use Full Hardware virtualization in our work, which is the virtualization of computer system as whole HW platforms, logical abstractions of its componentry, or just the functionality required to run various operating systems. Any software or operating system capable of running on the underlying physical hardware can be run in a virtualized environment (VM). Users can run many guest operating systems at the same time. In this case, the VM simulates enough underlying hardware to run an unmodified guest OS in isolation. Full virtualization provides the finest isolation and security for virtual machines (VMs), as well as making migration and portability easier. We are using a co-processor to enable Full Hardware virtualization and increase system performance by offloading processor-intensive operations from the main core processor. The main objective of the work is to utilize Full Virtualization technology to enable secured multi-threading and efficient resource utilization among the VMs using a single hardware i.e., the Co-processor. This is done by allowing separation of resources on 4KB boundary between VMs and designing a hardware block logic inside the co-processor which helps the hypervisor to lock down the data and control path to avoid VMs from impacting one another. We are also adding virtualized interrupts and thus reducing interrupt latency time to process interrupt. We observe in this work that Co-processor in a virtualized environment is reducing CPU overhead and in turn AHB bus bandwidth reduction. As this is hardware controlled rather than software, thus we get better performance.