Reviewer: Christoph Jechlitschek
Date: 9-15-2005
How would you rate this paper, relative to others we have read? top 50%, but not top 25%
How would you rate your knowledge of the topic of this paper? novice
What problem or issue does the paper address? Why is it important?
The paper examines the previously observed difference of throughput between TCP and TFRC. TFRC was designed to replace TCP in applications that require a more steady throughput (than TCP) while maintaining the same average throughput. In their experiments the authors showed that the throughput of TFRC can be up to 20 times more or ten times less than the throughput of TCP under certain congestions. The authors explore in their paper the reason for that behavior, determine which factors cause those differences, and try to capture their findings in equations. This research is important because TFRC was designed to be TCP-friendly, but newer findings show huge differences in behavior. If more and more applications will use it, it is important to address and hopefully fix those issues.
What are the main contributions of the paper and why are they important?
The authors performed a large amount of experiments to show the difference between TCP and TFRC. This is rather important because those experiments show exactly under which conditions the behavior of TCP and TFRC diverges. They describe their experiments in detail so that other researchers can repeat the experiments. The next contribution is the recognition of which factors cause the divergence. Finally the authors capture the behavior of both protocols in equations and use it to explain the differences.
How significant are these contributions relative to previous work?
Unfortunately, I'm not familiar with any previous work that was done.
Give detailed comments justifying your view of the paper.
While the paper doesn't propose any new method for addressing the differences found between both protocols, it helps to understand what causes them. So that paper should be seen as a first step of designing TFRC 2. The findings are well presented, however I was lost in their equations more than once. I had a hard time of choosing between the top 25% and the top 50%, it is definitely a good paper but it found nothing exceptional in it, therefore my choice.