[Defense] Hierarchical Real-Time Scheduling in Virtualized Systems: Design, Implementation and Optimization
Friday, May 14, 2021
3:00 pm - 4:00 pm
Pavan Kumar Paluri
will defend his proposal
Hierarchical Real-Time Scheduling in Virtualized Systems: Design, Implementation and Optimization
Virtualization in recent times has become an increasingly popular software technique with the advent of cloud and an ever-increasing need to develop and deploy a multitude of cloud-native applications. These applications take up a significant share of today’s world wide web, the majority of which implicitly require soft-real-time guarantees when hosted on servers at various data centers across the globe. In addition to this, Virtualization in embedded systems is also gaining popularity in the fields of aerospace and automobile research. The virtualized applications catering to the needs of aerospace and automobile industries however demand hard-real-time guarantees. To satisfy both the hard and soft real-time guarantees, the Hierarchical Real-Time Scheduling (HiRTS) framework specifies real-time scheduling policies to applications and also the Virtual Machines (VMs) hosting these applications. Conforming to the policies of HiRTS framework, a Regularity-based Resource Partitioning (RRP) model has been proposed that introduces the concept of regularity to provide a near-ideal CPU resource supply to all its VMs. Although, RRP model has been an active research topic with numerous theoretical contributions since its proposal, it is still devoid of a realistic test-bed across different platforms. A practical implementation of this model is therefore exigent in validating the claims established via theoretical contributions and also in providing a test-bed for future research in RRP. In addition, there is still a large scope for improvement in terms of CPU resource utilization. Besides, the absence of a robust fault-tolerance solution makes VMs operating under RRP model susceptible to faults thereby causing a complete collapse of one or more VMs. This proposal investigates into all the aforementioned problems and aims at providing a novel step-wise solution to each and every problem.
3:00PM - 5:00PM CT
Online via MS Teams
Dr. Albert M.K. Cheng, dissertation advisor
Faculty, students and the general public are invited.