CS848 Paper Review Form - Fall 2006

Paper Title: Dynamic placement for clustered web applications

Author(s):A. Karve, T. Kimbrel, G. Pacifici, M. Spreitzer, M Steinder, M Sviridenko, and A. Tantawi.

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
 [ ] Very well written
 [x] 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)

The paper addresses the problem of application instance placement in a distributed
environment based on load dependent and load independent resource demand
for applications. A placement controller that minimizes application movement across nodes
and strives for a load that is evenly distributed accross nodes is devised.
The placement problem is solved as a two-dimensional packing problem.
The paper discusses the residual, incremental, and rebalancing placements as well.
In addition, to improve effectiveness of the algorithm, a load reduction 
and multiple-runs algorithm variants are discussed.

The paper adds value to previous research and has identified its shortcommings.
For these reasons and more, I recommend the paper to be accepted.

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

S1. Application load is more balanced accross machines
S2. Three algorithm variants proposed and used

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

W1. Uses only memory and CPU as load independent and load dependent requirements respectively.
W2. Estimates for memory only an upper limit
W3. CPU is equalized across all nodes, which may be heterogeneous

10) Detailed comments for authors.

How can the placement changes be minimized when the memory load and CPU load is high?