- - - - - - - - - - - - - - - Part 1 - - - - - - - - - - - - - - - This part is confidential - for program committee only Reviewer (Please fill in) ======== Name: Keith Marzullo Affiliation: University of California, San Diego Paper (Please fill in) ===== Number: xxxxxx Title: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx... Authors: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Category: Regular Paper Rating (Please rate the following by entering a number between ====== 1 and 5 according to the meanings given below) Interest of the SUBJECT - not the paper - to XXXX: --------> 4 (1 = low ... 5 = high) Confidence in your REVIEW: --------------------------------> 4 (1 = low ... 5 = high) Contribution of the paper: --------------------------------> 4 (1 = low ... 5 = high) Technical content: ----------------------------------------> 5 (1 = poor ... 5 = good) Presentation style: ---------------------------------------> 3 (1 = poor ... 5 = good) Overall recommendation ====================== Please enter a number (meanings shown below) here: --------> 4 (1 = Strong reject, 2 = Weak reject, 3 = Undecided, 4 = Weak accept, 5 = Strong accept) Please provide a three-line summary of your reasons for the recommendation above (it will be seen by the entire program committee). This summary is REQUIRED. Three Line Summary: Pretty result about how to increase read or write availability in pessimistic syntactic quorum systems (by far the most common type). A perfect paper for DISC (nee WDAG) and PODC, a bit off-area for xxxx but not critically so. The result is fundamental: much future research in this area will need to refer to this paper. - - - - - - - - - - - - - - - Part 2 - - - - - - - - - - - - - - - Information to the author(s) Paper (Please fill in) ===== Number: xxxxxx Title: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Authors: xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Category: Regular Paper Summary comments for author(s) (Please enter a number representing ============================== the most appropriate choice) Subject matter: -------------------------------------------> 1 (1 = theory, 2 = paper study, 3 = experiment, 4 = practice) Interest of the subject - not the paper - to xxxx: --------> 4 (1 = low ... 5 = high) Novelty of the problem: -----------------------------------> 1 (1 = well known problem, 2 = new formulation of existing problem, 3 = new problem of limited importance to FT field, 4 = new important problem in FT field) Originality of the solution: ------------------------------> 3 (1 = minor originality, 2 = clever use of existing solutions and techniques, 3 = improvement of existing solutions, 4 = very innovative) Expected impact of results in the paper: ------------------> 2 (1 = minor impact, 2 = impact on small special area within FT field, 3 = impact on large area) Technically sound: ----------------------------------------> 1 (1 = yes, 2 = no, 3 = appears to be) Mathematics are essential, complete and correct: ----------> 1 (1 = yes, 2 = no, 3 = appears to be, 4 = does not apply) Self-contained: -------------------------------------------> 1 (1 = yes, 2 = no) Length meets submission requirements: ---------------------> 1 (1 = yes, 2 = no) Adequate reference to prior work: -------------------------> 1 (1 = yes, 2 = no) English is satisfactory: ----------------------------------> 1 (1 = yes, 2 = no) Written comments for author(s) ============================== Please enter your reasons for acceptance or rejection and suggest improvements to the paper. Reviews lacking informative comments will be disregarded. Comments: Presents two pretty techniques, built on the same base theory, of increasing either read or write availability for pessimistic syntactic quorum-based systems (by far the most commonly studied technique). XXXX has dealt with this topic, and it's a perfect paper for DISC (nee WDAG) and PODC. The results are fundamental, in that any future research in quorum-based systems will need to at least mention this paper. Most of my comments below are nits. Page 3, line 9: "suited replica" should be either "well-suited replica" or "suitable replica". The same holds for three lines down. Page 4, Section 3.1, line 3: citation [4] defines "fail-stop", not "fail-safe". You are using "fail-safe", in that you are not assuming a perfect failure detector. Page 5, line 1: You actually do assume a completely connected network (see page 10). I don't see why you make such a weak statement here; it's unnecessary. Page 5, lines 2-12: This is written in a very ugly style. Please don't use folded text like this. It is very hard (even harder) to read (to understand). Page 5, Section 3.2, lines 6 and 9, latent Romancism: the pronoun "he" should be "it". Page 7, lines -7 - -10: "The concept of closed coteries is a special case of the concept of acceptance sets..." I didn't understand this argument. You should either expand it so that it is understandable or drop it. Page 7, lines -5 - -6: I don't understand why closed coteries as write quorum sets are required to guarantee that write operations are performed under mutual exclusion. Doesn't the same hold for non-closed coteries for write quorum sets? Page 7, lines -1 - -3, a bunch of typos: why is RCP subscripted by A? What is Q_L? What is the difference between Q_W and Q_w? Page 8, Definition 4.3: this is very poor notation. You should be defining the members of A bar, not a single A bar that is the union of a bunch of sets. Page 9, proof of Theorem 4.2, the first word should be "It" rather than "Is". Page 10, Equation 5 (and those following from Equation 5): I think that this holds only for closed Q_{op}. Consider M = {a, b} and Q_R = {{a}, {b}}. P_R should be 1 - (1-p_a)(1-p_b) = p_a + p_b - p_a p_b yet Equation 5 computes P_r to be p_a + p_b - 2 p_a p_b. If Q_R is closed (that it, it includes {a, b} then Equation 5 computes the correct value. A simple explanation of why Equation 5 holds would probably point out this requirement. Page 10, line -4 (and at several other places in the manuscript): "is in point (1/2, 1/2) symmetric..." should be "is at point (1/2, 1/2) symmetric..." Page 12, Section 5.1, first word "Next" is not needed. Page 16, line 3, "necessarily be meet" should be "necessarily be met". - - - - - - - - - - - - - End of review - - - - - - - - - - - - - -