Dissertation/Thesis Abstract

Software Deployment Methods to Support Heterogeneous-ISA Application Migration
by Xing, Tong, Master's, Stevens Institute of Technology, 2020, 92; 27957383
Abstract (Summary)

Edge computing is a recent computing paradigm that brings Cloud services closer to the user. The most significant feature of Edge computing is that it enables low latencies between services and users. This is because Edge nodes are physically closer to a user. However, this advantage can be reduced if a user physically moves away from a specific Edge server. In order to provide a stable low latency service, services that run on Edge need to move among Edge nodes along with a user to maintain minimal physical distance. Different from the Cloud, Edge nodes usually are built with CPUs of different Instruction Set Architectures (ISA); therefore, homogeneous software migration – solution widely adopted in Cloud data-centers, cannot be used at the Edge.

This thesis introduces H-Container and its integration with Docker. H-Container enables a containerized natively compiled application to migrate across Edge nodes built with different ISA CPUs. Supporting Docker, allows heterogeneous-ISA migration with Docker. H-Container is a) easily deployable – implemented in user space, without any OS or hypervisor dependency; b) highly compatible – H-Container doesn’t need the source code of the application and c) Linux friendly – H-Container targets Linux, it can migrate most Linux application binaries.

Experiments demonstrate that H-Container adds no overhead on the application during the execution, while 10ms to 100ms added during the migration, and most importantly, in the experiment that is based on a real scenario, the output shows H-Container provides up to 94% increase in Redis server throughput when unlocking heterogeneity.

Indexing (document details)
Advisor: Xu, Jun, Barbalace, Antonio
School: Stevens Institute of Technology
Department: Computer Science
School Location: United States -- New Jersey
Source: MAI 82/1(E), Masters Abstracts International
Subjects: Computer science
Keywords: Criu, Docker, H-Container, Heterogeneous, Migration
Publication Number: 27957383
ISBN: 9798662407589
Copyright © 2021 ProQuest LLC. All rights reserved. Terms and Conditions Privacy Policy Cookie Policy