Dynamic Real-Time Virtualization and Cloud Computing
Abstract
Statement of Work:Current shared computing platform (cloud) architectures were derived from architecture targeted for time-insensitiveweb-service applications. As such we have witness its inadequacy in guaranteeing strict timing requirements of realtimeapplications. Kludges have to be made using side-channel (off-band) timing sources. Even with these kludgeswhich do not address the inadequacy of the underlying virtualization architecture, the resulting timing guarantee &accuracy (jitter) is yet to be desired for real-time applications.The proposed work aims to extend the PIs previous ONR funded research results for introducing real-time schedulinginto virtualized environment with four new integrated research thrusts that will substantially expand the scope of realtimevirtualization research and create new capabilities in real-time virtualization and cloud system technologies: (1)cache-aware compositional scheduling theory for real-time virtualization that incorporates realistic cache and systemoverheads on modern multicore platforms; (2) design of multi-mode virtual machines that can address mixed criticalitylevels; (3) a real-time cloud management architecture based on real-time virtualization and scheduling, includingsupport for real-time-aware migration of virtual machines; and (4) implementation, experimenal evaluation, opensourcerelease, and integration with widely used virtualization environments such as Xen.Objective:The PIs initial work on real-time virtualization platforms an compositional scheduling techniques has established keyfoundations towards meeting those challenges, and has followed a rapid transition from research results topublications in top-tier research conferences and journals and to open-source release within widely used virtualizationenvironments such as Xen. However many crucial research problems still remain open, and the Pis will address thosefurther challenges through a new integrated research plan that combines (1) new advances in compositional real-timescheduling theory to deal with cache effects and other overheads encountered by real-time virtualization platforms inpractice, (2) new multi-mode, dynamic, and adaptive resource management capabilities for real-time virtualizationplatforms, and (3) new infrastructure for managing real-time computing clouds based on those new dynamic real-timevirtualization platforms.Approach:The proposed work is characterized by the tight integration of theoretical and systems research. The proposedsystems research activities will provide a deployment platform within which to implement and evaluate experimentsbased on compositional schedulability analysis, an active area of research in real-time systems which until now hasremained largely theoretical, giving researchers a new means for practical evaluation of their research results and aclear technology transfer path. The proposed theoretical research also offers opportunities to strengthen thecapabilities of the real-time virtualization platform.This project is expected to make distinct contributions to the theory and practice of distributed real-time systemsdesign, analysis, development, and evaluation, which will substantially expand the scope of real-time virtualizationresearch and create new capabilities in real-time virtualization and cloud system technologies in the particular areas:??? Cache-aware compositional scheduling theory for real-time virtualization that incorporate realistic cache and systemoverheads on modern multicore platforms.??? Real-time virtualization and cloud technologies for multi-mode virtual machines with mixed criticality levels.??? Dynamic RT-OpenStack, a real-time cloud management architecture for distributed real-time systems based on realtimevirtualization and scheduling, which supports real-time-aware VM migration.??? Implementation and experimental evaluation using task sets derived from diverse real-time system scenarios, togauge both how accurately the proposed theoretical adv
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 26, 2018
- Source ID
- N000141612108
Entities
People
- Chenyang Lu
Organizations
- Office of Naval Research
- United States Navy
- Washington University in St. Louis