Enhancing end-to-end availability and performance by leveraging Internet redundancy.
Han, Junghee
2004
Abstract
In this thesis, we explore an emerging area of research---enhancing end-to-end performance and availability by leveraging the redundancy of the Internet. As the Internet is increasingly utilized as a business necessity, connectivity and network availability are becoming business-critical resources since even short-lived failures can generate significant losses. While ensuring a high level of end-to-end availability and performance is critical, current routing protocols have not been successful in masking path failures and/or network congestion to endhosts. This thesis presents the results of our work to maximize end-to-end application performance and availability by effectively leveraging the inherent redundancy of the Internet infrastructure. We first characterize the Internet's topological behavior, and study the effectiveness of two widely used existing architectures: multi-homing and overlay networks. In particular, this work quantifies the extent to which the redundancy of the Internet is utilized, and identifies several problematic features of current multi-homing and overlay networks. This study will help researchers and operators to gain insights into the realistic behavior of multi-homing and overlay networks and use these insights to develop better network design and configuration. Inspired by these observations, we propose a framework for topology-aware overlay networks by incorporating topology considerations in designing overlay architectures. In this framework, we develop several heuristics for intelligent node placement by which we can ensure that each overlay node is as topologically and geographically diverse as possible. On top of this architecture, we adopt a single-hop routing mechanism for better scalability. This thesis presents measurement-based verification to conclude that the proposed single-hop overlay routing performs as well as multi-hop routing with respect to both availability and performance. Finally, we develop the architecture for FVPNs (Fault-tolerant Virtual Private Networks) as an example application. The main idea of this work is to pre-calculate and install a set of redundant, minimally overlapping paths between endpoints that can be used as backup routes when primary paths fail. With FVPNs, we would be able to provide seamless network fail-over to VPN clients, independent of underlying IP networks. Overall, we believe that this study will help researchers and operators to understand the behavior of currently existing multi-homing and overlay networks, and give them some guidelines for designing and configuring the networks for better availability and performance.Subjects
Availability End Enhancing Internet Leveraging Performance Redundancy Virtual Private Networks
Types
Thesis
Metadata
Show full item recordCollections
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.