Show simple item record

dc.contributor.authorLiu, Yunpeng
dc.date.accessioned2014-05-22 17:52:52 (GMT)
dc.date.available2014-05-22 17:52:52 (GMT)
dc.date.issued2014-05-22
dc.date.submitted2014
dc.identifier.urihttp://hdl.handle.net/10012/8495
dc.description.abstractMost datacenter network (DCN) designs focus on maximizing bisection bandwidth rather than minimizing server-to-server latency. They are, therefore, ill-suited for important latency-sensitive applications, such as high performance computing, realtime analytic systems and high-frequency financial trading. Although there are a number of existing approaches to reduce network latency, they are only partially effective, workload dependent, and often require network protocol changes. In this thesis, we explore architectural approaches to building a low-latency DCN and introduce Quartz, a new optical design element consisting of a full mesh of switches connected by an optical ring. We can reduce the network latency of a hierarchical or random network by replacing portions of it with a Quartz ring. Our analysis shows that, in a standard 3-tier DCN, replacing high port-count core switches with Quartz can significantly reduce switching delays, and replacing groups of top-of-rack and aggregation switches with Quartz can significantly reduce congestion-related delays from cross-traffic. We overcome the complexity of wiring a complete mesh by using low-cost optical multiplexers that enable us to efficiently implement a logical mesh as a physical ring. We evaluate our performance using both simulations and a small working prototype. Our evaluation results confirm our analysis, and demonstrate that it is possible to build low-latency DCNs using inexpensive commodity elements without significant concessions to cost, scalability, or wiring complexity.en
dc.language.isoenen
dc.publisherUniversity of Waterlooen
dc.titleLeveraging Commodity Photonics to Reduce Datacenter Network Latencyen
dc.typeMaster Thesisen
dc.pendingfalse
dc.subject.programComputer Scienceen
uws-etd.degree.departmentSchool of Computer Scienceen
uws-etd.degreeMaster of Mathematicsen
uws.typeOfResourceTexten
uws.peerReviewStatusUnrevieweden
uws.scholarLevelGraduateen


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record


UWSpace

University of Waterloo Library
200 University Avenue West
Waterloo, Ontario, Canada N2L 3G1
519 888 4883

All items in UWSpace are protected by copyright, with all rights reserved.

DSpace software

Service outages