Referee 1 *******************************************

This paper describes design and implementation of CORBA CoG that provides
CORBA-based access to Grid services. It is very relevant and interesting
subject, as it has potential for constructing Grid Computing Environments
using (distributed) object-oriented approach. The functionality of CORBA CoG
is similar to that of Java CoG, and therefore can be regarded as an
alternative to Java-based grid middleware.

Integration of Globus and CORBA is not an easy task, since (as authors
carefully pointed it out) there is a significant overlap in architecture of
the two systems. The critical element is as usual the security.
Unfortunatelly, the paper does not describe this in full details. A hybrid
solution is used: CORBA security services and GSI. An itermediary object is
used to "forward" the client credentials to the GRAM server. What is not
explained is how the client authenticates to the intermediary, and how the
user credentials (a proxy certificate) is stored.

However, this and some other small glitches does not change my general
impression that the paper descibes a very interesting and good work (which
hopefully will be continued), and the paper itself is very well written.

Definitely I recommend this paper for publication in the special issue of
concurrency.

Referee 2 *******************************************

Typos & some suggestions:
Section 1: remote ' remo te
Page 14 ' 
-- : the document was probably typeset in latex probably a \ is missing. Should be just a
dash.
"its" instead of "it s"
Since there are a lot of acronyms/abbreviations in the paper, the authors need to double
check if they have been expanded prior to their use. Eg. GSI /GRAM on page 6 both of
which are expanded on page 7.
Page 15: reference to Figure 7 is followed by two periods.
Figure  8 is referenced as Figure 88

Concerning Figures 5,6,7 it would help if they are "closer" to the places where they are
referenced. The circles in 6,7 could be improved.

Technical queries
Handling Grid Security in CORBA:
*	I would be interested in knowing the CA structure that has been proposed within the
overall framework. Is there a CA structure similar to the X.509 standard, which has a
hierarchical model for certification authorities. Is it possible to authenticate beyond
the realms of individual CA's since there is probably no root CA. Are Certificate
Revocation Lists supported?

*	How long are the certificates valid for,  the duration of the session, a few days
etc.

*	Either due to space constraints or due to some other reasons there seemed to be
little discussion on the security process. My reading gave me the feeling that deploying
certificates was the panacea.

The authors have discussed the benefits of the Grid/CORBA combination. CORBA and Grid
service both come with their own set of problems. The merger problem is not fully solved
unless the conundrum involving drawbacks involving CORBA, Grid and their combination are
fully addressed. Some drawbacks could be amplified, some eliminated - if that is the
case, what are those. The combination itself would have a roadblock or two to scale, how
are they addressed.

A clear discussion on the type of communication libraries that need to be present at
clients accessing such a merged model would be very interesting.

The paper would be interesting once these issues are addressed.


Referee 3 *******************************************

This paper discusses a research project to design and deploy a CORBA-based
commodity software system that works on the grid.  The idea is to allow
application developers that have written their applications within a CORBA
framework to be able to utilize the CoG Kit and enable their applications
on the grid.

While I am not a big fan of CORBA per se, it is my opinion that this is
exactly the type of toolkits that are necessary to make the grid useful for
"real" application developers.  Currently, there is a LARGE gap people
middleware developers, that develop tools such as Globus and Legion, and
"real" application developers that have large, complex applications.
Tools, such as the CoG Kit, are very much in need if Grid computing is to
be successful to the application community.

The paper is well written and has some tables that CORBA applications
programmers will find useful.tational Grid?