CS848 Paper Review Form - Fall 2006
Paper Title: Xen and The art of virtualization

Author(s): P. Barham, B. Dragovic, K. Fraser, S. Hand, T. Harris

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
 [X] Very good (comparable to best conference papers)
 [ ] 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)

The main contribution of this paper is that it brings forward the major issues with 
virtualization in the x86 architecture in an attempt to objectively address them. 
It explains in great detail a paravirtualized solution ("Xen") and provides sound 
arguments to justify the design choices made in its implementation. 

While the paper may not be full of new ideas, it provides a very comprehensive 
discussion of the various virtualization subsystems and offers as a bechmarked 
implementation  (near optimial in some cases) of x86 virtualization. The paper 
also acknowledges several drawbacks in the paravirtualization approach which opens 
the arena for future work. I would accept this paper on the grounds that publishing 
it would will catalyze more research in this field from both the software and hardware 
areas.


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

S1 - Clearly identifies how (by benchmarking) and why (by discussing design details)
paravirtualization outperforms full virtualization.

S2 - Provides excellent scalability by minimizing the hypervisor's state footprint 
to about 20kB.


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

W1 - The most obvious (and acknowledged) shortcoming was the need to modify guest os code.

W2 - The paper does not make clear what the distinction between a guest OS kernel and the 
hypervisor is. How closely knit do they have to be for their proposed paravirtualization 
technique? Will the hypervisor and the guest kernel form a newer amalgamated/abstracted kernel?


10) Detailed comments for authors.

Some parts of the paper were led into without mentioning the need to address them. For example
Time and timers (3.3.2) was not necessary in this paper.

One particular phrase i saw throughout the paper was "the guest OS may" and it was often not clear
whether the feature/problem being discussed was an inherint part of the guest operating system(s) or was
being induced by applications in the Guest OS.