Consolidation and isolation are key technologies that promoted the undisputed popularity of virtualization in most of the computer industry. This popularity has recently led to a growing interest in real-time virtualization, making this technology enter the real-time system market. However, it has several issues due to the strict timing guarantees contracted. Moreover supporting legacy software stacks adds another level of complexity when the software is a black box. We present KairosVM, a latency-bounded, real-time extension to Linux's KVM module. It aims to bridge the lack of communication of the real-time requirements between the guest scheduler and the host scheduler, exploiting virtual machine introspection. The hypervisor captures the real-time requirements of the guest by catching previously added undefined instructions, without the need to do any modification to the guests. Our evaluations show that KairosVM's overhead is negligible when compared to existing introspection solutions thus can be used in real-time.
This paper describes a new hypervisor built to run Linux in a virtual machine. This hypervisor is built inside Anaxagoros, a real-time microkernel designed to execute safely hard realtime and non real-time tasks. This allows the execution of hard real-time tasks in parallel with Linux virtual machines without interfering with the execution of the real-time tasks.We implemented this hypervisor and compared performances with other virtualization techniques. Our hypervisor does not yet provide high performance but gives correct results and we believe the design is solid enough to guarantee solid performances with its future implementation.
We present PrVM, a framework for scheduling real-time VMs on multicore hardware. It addresses the intersection of the following problems: probabilistic real-time scheduling, VM scheduling, and full virtualization. Though each of these problems have been studied, their intersection - motivated by the need to consolidate multiple real-time software stacks, whose applications can be defined via probabilistic timing properties, onto a single embedded platform - is empty. PrVM uses a probabilistic model and timeliness optimality criterion. PrVM schedules VMs as server-like processes, computes time budgets using probabilistic methods, and aggregates task time budgets into VM time budgets. Experimental evaluations, using simulations and a concrete implementation, confirm the framework's effectiveness for synthetic benchmarks and multimedia applications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.