Identifier

etd-07102013-180431

Degree

Doctor of Philosophy (PhD)

Department

Computer Science

Document Type

Dissertation

Abstract

Internet traffic is expected to grow phenomenally over the next five to ten years. To cope with such large traffic volumes, high-speed networks are expected to scale to capacities of terabits-per-second and beyond. Increasing the role of optics for packet forwarding and transmission inside the high-speed networks seems to be the most promising way to accomplish this capacity scaling. Unfortunately, unlike electronic memory, it remains a formidable challenge to build even a few dozen packets of integrated all-optical buffers. On the other hand, many high-speed networks depend on the TCP/IP protocol for reliability which is typically implemented in software and is sensitive to buffer size. For example, TCP requires a buffer size of bandwidth delay product in switches/routers to maintain nearly 100\% link utilization. Otherwise, the performance will be much downgraded. But such large buffer will challenge hardware design and power consumption, and will generate queuing delay and jitter which again cause problems. Therefore, improve TCP performance over tiny buffered high-speed networks is a top priority. This dissertation studies the TCP performance in 10Gbps high-speed networks. First, a 10Gbps reconfigurable optical networking testbed is developed as a research environment. Second, a 10Gbps traffic sniffing tool is developed for measuring and analyzing TCP performance. New expressions for evaluating TCP loss synchronization are presented by carefully examining the congestion events of TCP. Based on observation, two basic reasons that cause performance problems are studied. We find that minimize TCP loss synchronization and reduce flow burstiness impact are critical keys to improve TCP performance in tiny buffered networks. Finally, we present a new TCP protocol called Multi-Channel TCP and a new congestion control algorithm called Desynchronized Multi-Channel TCP (DMCTCP). Our algorithm implementation takes advantage of a potential parallelism from the Multi-Path TCP in Linux. Over an emulated 10Gbps network ruled by routers with only a few dozen packets of buffers, our experimental results confirm that bottleneck link utilization can be much better improved by DMCTCP than by many other TCP variants. Our study is a new step towards the deployment of optical packet switching/routing networks.

Date

2013

Document Availability at the Time of Submission

Secure the entire work for patent and/or proprietary purposes for a period of one year. Student has submitted appropriate documentation which states: During this period the copyright owner also agrees not to exercise her/his ownership rights, including public use in works, without prior authorization from LSU. At the end of the one year period, either we or LSU may request an automatic extension for one additional year. At the end of the one year secure period (or its extension, if such is requested), the work will be released for access worldwide.

Committee Chair

Park, Seung-Jong

Share

COinS