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)

This paper designs and implements a ¡°placement controller¡± that can dynamically allocate computing resources to a web application cluster by adjusting the size and placement of application instances based on user defined policies. The algorithm behind the controller can compute placement based on load-dependent and load-independent capacity constraints. In order to reduce the placement overhead, the placement algorithm tries to minimize the number of placement. The paper also proposes three simple but effective variants to the placement algorithm that can be used in different circumstance. 

In general the paper is well-written. The experiments are empirical, and the results clearly show the placement algorithm is effective. 


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

S1: The placement problem is NP hard, but the placement algorithm presented in this paper uses a different way to find a good enough placement solution instead of solving the NP hard problem directly.

S2: The paper provides three variants to the placement algorithm that can be used in different circumstance. 


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

W1: The paper didn¡¯t address the circumstance when the entire system is overloaded, how the placement algorithm should behave. 


10) Detailed comments for authors.

A few comment about W1 above. When the entire system is overloaded, ie the application can¡¯t satisfy the demand, some applications have to give their computing resources to other applications based on how important they are. Therefore it is good to have a priority value defined for each application in the system. The placement algorithm should take that into account when the entire system is overloaded.