Home > Events > Oral Candidacy - KangKang Li

Oral Candidacy - KangKang Li

Start: 10/3/2017 at 2:30PM
End: 10/3/2017 at 5:30PM
Location: 315 Stinson Remick
Attendees: Faculty and students are welcome to attend the presentation portion of the defense.
Add to calendar:
iCal vCal

KangKang Li
Oral Candidacy
October 3, 2017   2:30 pm      315 Stinson
Adviser:  Dr. Jarek Nabrzyski
Committee Members:
Dr. Gregory Madey        Dr. Maciej Malawski        Dr. Scott Nestler

Title

"RESOURCE ALLOCATION STRATEGIES IN 3D TORUS-BASED HIGH

PERFORMANCE COMPUTING AND CLOUDLET MESH SYSTEMS"

Abstract

In the topic of resource allocation of computing systems, network topology plays an important role in the way that the jobs should be allocated. In this work, I address two resource allocation problems. The first one is the topology-aware job scheduling and placement problems in high performance computing (HPC) systems, where a 3D torus-based network topology is used. The second one is related to the resource allocation problem in edge cloud systems. To be specific, I study the virtual machine (VM) placement problem in the cloudlet mesh. A wireless mesh network topology and the adaptive bandwidth provided by the Software Defined Networking (SDN) technologies are applied in the considered cloudlet mesh architecture. To address the topology-aware job scheduling and placement problems on 3D torus-based HPC systems, I focus on reducing system fragmentation and improving system utilization. Firstly, from the job scheduling aspect of view, I propose a packing-based job scheduling strategy, which reduces the external fragmentation by using the First Come First Served (FCFS) + backfilling strategy. Secondly, I study the first case of job placement problem, where each job is allocated a convex prism shape. I propose a job placement algorithm based on a local and a global migration processes, which aims at reducing the internal and external fragmentation during the job placement process. Thirdly, I study the second case of job placement problem,