8980 Critique
<your name>

<paper title>
<authors>

1. Positive Points 

[What was unique and interesting about this paper.  Provide at least two bullet items.]
- Item 1
- Item 2

2. Negative Points 
[What was limiting or a weakness of this paper.  Provide at least two bullet items.]
- Item 1
- Item 2


You are welcome to provide more items. I would prefer a fewer number
of deep and well-thought out items than a bunch of shallow ones.  At
least 1 item in each category, cannot be cited in the abstract or
conclusion, i.e. it must be your independent analysis, and at least
one item must relate to another paper or system (not cited in the
paper itself). Examples of appropriate and inappropriate
(hypothetical) answers:

Good ones:

Liked: The CandyCane Computing Cloud provided bounded statistical
guarantees on compute and network performance. This broadens the scope
of applications appropriate for clouds. It was reminiscent of how QoS
was achieved in the FishScaler multimedia system but also included the
clever idea of user-provided feedback to adjust allocations.

Disliked: The CandyCane Computing Cloud has a huge security hole. For
example, the exposed root node could be subject to a DDOS attack. The
StarBurst Cloud does not have this problem due to its virtualization
layer. The CCC also had a significantly more complicated interface
than EC2 - 25 vs. 5 functions, many of which could have collapsed. For
example, ...


Not so good:

Liked: The CandyCane Computing Cloud was cheaper than Amazon's EC2 and
seemed to be much faster.  
  [problem: need to say why it was faster and cite evidence ... cheaper is not a strong technical argument]

Disliked: The CandyCane Computing Cloud could not handle workflow
applications nor MPI applications similar to most other clouds.
  [problem: limitation not unique to CandyCane]