Show simple item record

dc.contributor.advisorHari Balakrishnan.en_US
dc.contributor.authorIwashima, Hiroyoshi, 1980-en_US
dc.contributor.otherMassachusetts Institute of Technology. Dept. of Electrical Engineering and Computer Science.en_US
dc.date.accessioned2005-09-26T20:36:23Z
dc.date.available2005-09-26T20:36:23Z
dc.date.copyright2003en_US
dc.date.issued2003en_US
dc.identifier.urihttp://hdl.handle.net/1721.1/28464
dc.descriptionThesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2003.en_US
dc.descriptionIncludes bibliographical references (p. 65-66).en_US
dc.description.abstractThis thesis presents and evaluates the design and implementation of a user-level library that performs bandwidth allocation among multiple application flows using multiple TCP connections. This work is motivated by emerging trends in network overlay applications to send multiple flows of data between machines. Existing approaches to manage the network either do not offer the control over network resources that these applications need, or are difficult to adopt. We present a novel user-level approach to this problem that uses multiple TCP connections, and we also present an evaluation of multiplexing strategies for the scheduler that multiplexes the multiple application flows onto the multiple TCP connections. We implemented and designed two scheduling algorithms, striping and pinning, to perform the multiplexing and evaluated them for various application behavior and various link characteristics through emulation. We describe how striping is preferred for applications that are not delay dependent, as well as applications that are not order dependent, while pinning is preferred for applications that are heavily memory constrained. Our results show that, for delay-constrained applications, links with low loss and low number of cross traffic favor the striping scheduler, while high loss or high numbers of cross traffic favor the pinning scheduler. Delay-constrained applications that maintain high number of flows also favor the pinning scheduler, while those that do not favor the striping scheduler. We conclude that a user-level approach is both feasible and preferred, and that both multiplexing approaches are viable options based on network characteristics and application behavior.en_US
dc.description.statementofresponsibilityby Hiroyoshi Iwashima.en_US
dc.format.extent66 p.en_US
dc.format.extent3876583 bytes
dc.format.extent3882882 bytes
dc.format.mimetypeapplication/pdf
dc.format.mimetypeapplication/pdf
dc.language.isoen_US
dc.publisherMassachusetts Institute of Technologyen_US
dc.rightsM.I.T. theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission. See provided URL for inquiries about permission.en_US
dc.rights.urihttp://dspace.mit.edu/handle/1721.1/7582
dc.subjectElectrical Engineering and Computer Science.en_US
dc.titleDifferential bandwidth allocation with multiplexed TCP connectionsen_US
dc.typeThesisen_US
dc.description.degreeM.Eng.en_US
dc.contributor.departmentMassachusetts Institute of Technology. Department of Electrical Engineering and Computer Science
dc.identifier.oclc57125580en_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record