Review 1 Familiarity Recommendation Familiar (I am well aware of research work in this topic) (3) Likely accept (top 20% but not top 10%, significant contribution) (4) Contributions (What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper.) The paper develops and investigates a strategy for the co-existence of lossy flows with delay-based flows within TCP. Strengths (What are the major reasons to accept the paper?) The paper is well done, the problem stated is clear, the evaluations done analytically are very sound, and the emphasis on broad discussions reveals good results. Weaknesses (What are the major reasons NOT to accept the paper?) The use of PERT is ok, but it would see the need to be mentioned in the abstract, paper title, and Sec 1 (which happens here exactly once) - otherwise, "traditional" TCP seems to be used to see more "applicable" results? In which way would the Internet behave, if PERT and other mechanisms would run in "parallel"? Detailed Comments (Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors.) Address weaknesses listed above Compress Sec 1 to the core infos and don't broaden your arguments too much. Related work - besides PERT - may be useful to be discussed, since functionality, maintenance and management aspects for the network, and advances in more clearer numerical form are beneficial. A clear simulation model and topology would help to improve Sec 2 by far. The set of simulation results presented are good! The scalability discussion seems to be an add-on only, isn't it? Sec 5 does NOT include any conclusions at all, it only covers a very weak summary. Any application choices or discussion of the usefulness of the scheme investigated are missing. Review 2 Familiarity Recommendation Some knowledge (I am marginally aware of research work in this topic) (2) Likely accept (top 20% but not top 10%, significant contribution) (4) Contributions (What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper.) This paper discusses the issue of incremental deployment of delay-based TCP variants in a mixed environment where loss based flows are present. It studies the ability of delay-based flows to switch to a loss-based mode in the presence of competing loss-based flows, and to switch back to delay-based mode when loss-based flows no longer compete with them. It shows that current schemes are not likely to perform such a switch back, while scheme proposed in the paper does. Strengths (What are the major reasons to accept the paper?) The paper is well written, provide some intuitive observation and claims to introduce the problem, which are rigorously backed up with analysis and evaluation through simulation Weaknesses (What are the major reasons NOT to accept the paper?) Section IV is a bit cryptic, but I can't give recommendation on how to make it easier to read. The section on the limitations of the scheme is a bit short, IMO. It should be expanded so that the reader can more easily figure out the impact of these limitations. Detailed Comments (Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors.) Section II. In beginning of Section 2, I read : "In particular, in the presence of many PERT flows, there is no reason to believe that the average queueing delay will reduce below this threshold once loss-based flows leave the network". You actually show evidence of this in section II.A. It might be appropriate to point to these evidence instead of this "no reason to believe" that is unnecessarily unrigorous. I would need more details on the g() function. Section III.B : Limitations You shoot yourself in the foot with Limitation (A). This limitation requires much more meat, i.e., an intuition to the reader of what you mean by "very large number of". The sceptic will say that you always have a very large number of flows... Also, if you find that this revert to loss-based behavior is a feature, rather than a limitation, in the presence of a "very large number" of flows, why describing this as a limitation rather than as a feature ? Limitation (B) deserves an explaination. There is still room in this paper, take advantage of it to discuss this with a § rather than with a single sentence. Limitation (C). Are you increasing the likelihood of being unfair as compared to unmodified mPERT ? I don't get Limitation (D). I guess I wouldn't be the only one not understanding this. Please expand Review 3 Familiarity Recommendation Some knowledge (I am marginally aware of research work in this topic) (2) Likely accept (top 20% but not top 10%, significant contribution) (4) Contributions (What are the major issues addressed in the paper? Do you consider them important? Comment on the novelty, creativity, impact, and technical depth in the paper.) The authors introduced a simple strategy to make the fair coexistence possible and to ensure that delay-based flows will revert back to the delay-based operation when loss-based flows are no longer present so that delay-based congestion control and loss-based congestion control can work together. The authors solve this problem by modifying the recent work PERT and refer the modification as mPERT. The works is very interesting. Strengths (What are the major reasons to accept the paper?) The paper is well written, and it does proposed a solution about how to make the coexistence of delay-based congestion control and loss-based congestion control. The paper provides stability proof. As shown in the emulation results and stability analyzing, the solution has good performance. Weaknesses (What are the major reasons NOT to accept the paper?) Simple extension to existing solutions, so the originality of the work is not outstanding. Detailed Comments (Please provide detailed comments that will help the TPC assess the paper and help provide feedback to the authors.) The study of coexistence of different kind of congestion control is very interesting. The authors do some modification to the PERT and use it to solve the coexistence problem. As shown in the emulation results, it works. However, there is a very small difference between the mPERT and the PERT. The mPERT can be treated an extension of PERT, which hurt the originality of the paper.