CS848 Paper Review Form - Fall 2006

Paper Title: Conflict-driven load control for the avoidance of data-contention
             thrashing

Author(s): Axel Moenkeberg and Gerhard Weikum

1) Is the paper technically correct?
 [X] Yes
 [ ] Mostly (minor flaws, but mostly solid)
 [ ] No

2) Originality
 [ ] Very good (very novel, trailblazing work)
 [X] Good 
 [ ] Marginal (very incremental)
 [ ] Poor (little or nothing that is new)

3) Technical Depth
 [ ] Very good (comparable to best conference papers)
 [X] Good (comparable to typical conference papers)
 [ ] Marginal depth
 [ ] Little or no depth

4) Impact/Significance
 [ ] Very significant
 [X] Significant
 [ ] Marginal significance.
 [ ] Little or no significance.

5) Presentation
 [X] Very well written
 [ ] Generally well written
 [ ] Readable
 [ ] Needs considerable work
 [ ] Unacceptably bad

6) Overall Rating
 [ ] Strong accept (very high quality)
 [X] Accept (high quality - would argue for acceptance)
 [ ] Weak Accept (marginal, willing to accept but wouldn't argue for it)
 [ ] Weak Reject (marginal, probably reject)
 [ ] Reject (would argue for rejection)

7) Summary of the paper's main contribution and rationale
   for your recommendation. (1-2 paragraphs)
This paper provides a good example of the cycle of doing various emperical studies 
in computer science: having a purpose, conduct experiments through which
theritical generalization is found and analyzed, do experiments afterwards for 
evaluations. The main contribution of the paper is in three folds as stated clearly 
in the paper: a simulation study of how to measure data contention, an algorithm for
fully automatic DBMS load control, and a simulation study of the algorithm's behavior 
and evaluations for various overload situations. The paper concerns only data-contention
thrashing, caused by excessive lock conflict, and argues with evidence that limiting 
conflict rate rather than the DMP yields satisfactory performance regardless of the 
transaction types in the workload. The approach provided is shown to perform well in
coping with overload, reacting to load peaks and most significantly, it react fast to
load changes. In short, the paper gets a criterion in measuring data contention, an
algorithm in automatic load control, and a good experimental evaluation/comparision.

The paper is very well written in that its goal and contributions are precisely and 
clearly stated; the line of thought is evidently presented; experiments for different
purposes are well designed and conducted; the algorithm and evaluations are well analyzed.

8) List 1-3 strengths of the paper.  (1-2 sentences each,
identified as S1, S2, S3.)

S1: Good empirical study
S2: well conducted experiments
S3: thorough considerations for extensive possibilities


9) List 1-3 weaknesses of the paper (1-2 sentences each,
identified as W1, W2, W3.)

W1: Does not rule out all possibilities and thus room for future improvements


10) Detailed comments for authors.
Maybe a bit more discussion about the problems with choosing the fewest locks held as
the deadlock victim and restart with no delay. The current discussion is not sufficient.