Theme of ResearchThe Internet has grown tremendously along several dimensions over the past two decades. Yet, the core transport mechanisms used in its operation have not changed much. In particular, after observing network congestion in the late 80s and early 90s, several congestion control mechanisms---instantiated mostly as different versions of the TCP protocol---were developed. These mechanisms are ubiquitously deployed today. Meanwhile, the Internet infrastructure has grown by several orders of magnitude, both in its structure as well as traffic-carrying capacity. It is difficult to blindly believe that the assumptions on which network designs were based a decade ago are valid even today. It is natural to ask:
- Is there any congestion in today's Internet? If so, where is it?
- Are assumptions used in the design of legacy transport mechanisms valid any more?
- How does the invalidity of assumptions impact past work?
- How can mechanisms and analysis techniques be redesigned after discarding invalid assumptions?.
Unfortunately, we do not know the answers to most of these fundamental questions. It is the goal of our research to answer these by:
- Developing measurement and analysis techniques that enable end-users to develop a fundamental understanding of the transport performance of networks and validate legacy assumptions.
- Studying the impact of invalid assumptions on the design, analysis, and evaluation of existing transport mechanisms.
- Designing new mechanisms and analysis techniques after discarding high-impact invalid assumptions.