Reviewer: Christoph Jechlitschek
Date: 9-29-2005
How would you rate this paper, relative to others we have read? bottom 50%
How would you rate your knowledge of the topic of this paper? familiar, but not expert
What problem or issue does the paper address? Why is it important?
The paper addresses the problem that TCP is not able to utilise all bandwidth on links with high bandwidth delay product. That's because individual flows scale back very fast on packet losses but recover bandwidth only very slowly because of the additive increase. On the other side, proposals that provide router support to address this issue incur large computational overhead or require additional field in the IP header that in general are not available.
What are the main contributions of the paper and why are they important?
The main contribution is the design of a ligthweight mechanism to provide router feedback in the two bit ECN fields. In addition with multiple increase(MI) when bandwidth is available and additive increase(AI) when the bottleneck link begins to saturate that mechanism results in higher link utilisation. A smaller contribution is the use the adaptive parameters for AI and MI that try to offset the penalty incurred by heterogenous roundtrip times.
How significant are these contributions relative to previous work?
While the authors propose a well designed system that works well in simulation and is supported by a fluid model, its significence is unfortunately rather small. That's because many similar systems exist, most of them which are not used due to the ossification of the Internet.
Give detailed comments justifying your view of the paper.
While the combination of lightweight router feedback and three different load regions (with different behavior in each) are fairly new, however the impact that this paper makes is rather small. The Internet is too ossified and there are too many other proposals out.