Computer Science Thesis Proposal
— 3:00pm
Location:
In Person and Virtual - ET
-
Reddy Conference Room, Gates Hillman 4405 and Zoom
Speaker:
ADITHYA ABRAHAM PHILIP
,
Ph.D. Student, Computer Science Department, Carnegie Mellon University
https://aphilip.cc/
Architecting the Matrix: Accurately Parameterizing Internet Performance Testing for Realistic Evaluation
The performance of Internet services -- be it file download completion times, video quality, or lag-free video conferencing -- is heavily influenced by network parameters. These include the bottleneck bandwidth, packet loss, network delays, and how fairly the bottleneck link is shared with other services. However, current techniques to evaluate service performance display three major issues: (a) overemphasis on the role of Congestion Control Algorithms (CCAs) in determining application performance; (b) testing predominantly in settings representing the "edge" of the Internet, and not the core; (c) testing in settings that do not necessarily reflect where congestion occurs on the Internet today.
The goal of this thesis is to improve the state of the art in testing for a more realistic evaluation of Internet service performance. We achieve this by changing measurement methodology to evaluate deployed Internet services as opposed to just their underlying CCAs, testing in more diverse network conditions, and identifying more realistic parameters for evaluations.
We build Prudentia, an Internet fairness watchdog, to understand how fairly two Internet services can share a bottleneck link. In addition to discovering extreme unfairness on the Internet today, we gain key insights into improving current testing methodology -- (a) The most and least fair services both use variants of the same CCA, highlighting the need to test services in addition to CCAs; (b) network settings can drastically affect even service-level fairness outcomes, necessitating their careful selection.
We then examine the changes in CCA behavior when evaluated in settings representing the core of the Internet as opposed to the edge. We find that the change to core Internet speeds and flow counts dramatically alters fairness outcomes, and challenges long-standing assumptions about CCA behavior. This highlights the need to run Internet evaluations in more diverse settings.
These discoveries raise a crucial question: do current Internet performance evaluation parameters accurately reflect real-world congestion scenarios? In the final part of this thesis, we propose to answer this by leveraging comprehensive end-to-end measurements from a leading video-streaming service to identify the prevalent network conditions of Internet congestion. This groundwork will allow the development of benchmarks that accurately reflect real-world Internet usage, ensuring future evaluations are relevant and reliable indicators of CCA and service performance.
Thesis Committee:
Justine Sherry (Chair)
Srinivasan Seshan
Theophilus A. Benson
Renata Teixeira (Netflix)
Additional Information
In Person and Zoom Participation. See announcement.