------------------------------------------------------------------------------

	Overall, again I found this to be a well written paper.  The
scores are discussed below:

IMPORT:
	This work is definitely important to the OS community as it
discusses a new type of operating system structure.  In addition, the
extensible operating systems seem to perform much better than the
monolithic operating systems.  (Also, I am assuming that
since we read the Exokernel paper for class, this must be a hot area).
Thus a 7 for import.  

NOVELTY:
	I am assigning a 4 as average to novelty because the paper seemed
to be mostly a summary of the different extensible operating systems.
Although, in the summaries there were some analyses of each extensible
operating systems.  However, I would have liked to see a section at the
end, outside of the conclusion, that tied things together.

QUALITY:
	Again this was a well written paper and the authors seemed to have
a clear understanding of their subject.  Of what critique was there, seemed
to be well founded.   So, since the critique that there was good but the fact 
that there could have been more critique is why I chose 6.


OVERALL:
	This paper gave a good overview of extensible operating systems.
The paper was structured well and was consistent in its format which made
it easy to read.  In addition, each extensible operating system was given a 
good summary.  However, I felt that while the first two summaries gave a 
good short analysis of the system, the other summaries just gave a summary 
of the system.  So, in other words, one suggestion would be to cut down 
the summaries for each section and spend an additional section or two 
reviewing such things as safety features and other similarities/contrasts.
Overall, this paper earned a score of 6 and should be accepted to this
conference.
------------------------------------------------------------------------------


This paper was well layed out.  The structure of the first few sections, 
introducing operating systems concepts, and purposes provided helpful 
information for the rest of the paper, and served as a guideline for 
systems examined. While the breakdown of the rest of the paper into 
systems studied was also a clear one, it perhaps left too much systhesizing 
for the reader.

Their coverage and discussion of the issues was thorough. The range of 
papers surveyed was ideal; they covered many mechanisms for
extending operating systems thoroughly (though is some places the 
mechanisms described were still a little unclear).  I also felt that the 
conclusions that they drew about the applicability of the systems studied 
were dead on, and showed a high level of critical thought. 

In the conclusion,
they pointed out some open problems in Extensible Operating Systems, and 
discussed some of the issues and feasible solutions, which was a great way
to close the paper. They also summarized and drew conclusions about the
systems they surveyed in the paper, however, perhaps they didn't quite 
do enough of this and spent too much time redeveloping application domains
for extensible operating systems. 

The distinction of operating systems services into saftey mechanisms and 
multiplexing of system resources was fundamentally helpful for this paper. 
The discussion of extension techniques was good, though a little chaotic. It
laid out the rest of the paper really well, though the bulletted list of
serurity problems seemed unreferenced in the rest of the paper. The small 
summaries at the end of each section are reasonably good and helpful, but they
vary in quality from really good to adequate. I felt that the ones that 
compared/contrasted systems were the best. 

I felt that the coverage of such a wide range of techniques was the best part 
of this paper. 

------------------------------------------------------------------------------

Overall, this group did a good job in surveying extensible 
operating systems and presenting their results. The focus of the
paper was clear and the amount of details show that
the group really read and understood the reference papers.

I gave a rating of 5 in Import and Novelty because it seemed that
the important features described were thoughtfully chosen from
each OS.

This may be nitpicky, but the word "inherent" was used often, and,
at times, some more detail should have been given to explain why
some object or action "inherently" possessed some attribute
(ie, section 2.2).

The Abstract was nicely written, though the concluding paragraph
of the Abstract could have been more concise. The argument stated
here was that extensible operating systems clearly give an advantage
to server applications rather than general purpose computing. This
argument could have simply been posed here in the Abstract by itself,
with the supporting evidence summarized in the Conclusions section.

The Conclusions section was well written, and gave a balanced view
of extensible operating systems by describing some of the
disadvantages they share. However, I had one question that partly
prevented me from giving this survey paper a 6 in Quality. It
was concerned with software engineering issues:

  In extensible operating systems, more hardware details are exposed
  to the application layer, and doesn't that increase the difficulty
  of code maintenance (and reuse)?

Even though Quality is the most important of the first three
categories, the Overall rating is actually a 6, reflecting my
opinion that this paper should be accepted by the program committee.
------------------------------------------------------------------------------
This paper gave a very concrete treatment of the subject, presenting 
performance results from a number of real systems.  This was
very good.  There were a few grammatical errors that detracted from my 
enjoyment of the paper.
------------------------------------------------------------------------------
(1) Not many new ideas/future research areas pointed out. Infact only 2 that I 
crossed upon. One was safety issues with regards to grafts which I believe 
has already been mentioned by the authors (according to whats written in the 
sections previously). The other was interaction of different extensions which 
was just pointed at briefly.

(2) Some observations/criticisms are too brief to understand & so appear less
convincing to the reader (atleast for me). 
------------------------------------------------------------------------------
The traditional monolithic operating system with its growing kernel
is known to be hard to develop and maintain. Moreover this structure
of the operating system itself puts a heavy burden of overhead on 
some kinds of applications. The new approach of the micro-kernel 
that is extensible by grafting other modules to do the services of
the traditional kernel. This survey provides an adequate summary of
the various techniques of extending the kernel. The paper puts a 
great emphasis on the security issue and thoroughly deals with this 
problem when presenting each individual technique. However except 
for the case of Exokernel that stresses the management of resources 
the authors do not discuss adequately how the models use the new
modules to provide the traditional services of an operating system.
Although the survey gives a good evaluation of the various techni-
ques and points out the problem of interaction of different exten-
sions in the conclusion, it fails to suggest a direction for further
research to solve this problem. But in general this is a good sum- 
mary of new techniques to solve an important issue in operating sys-
tem architecture.