First of all, I really appreciate having this survey available. Like others commented, this survey should be re-done on 
a regular basis, trying to capture effects of progress with Grid middleware.

The current document, however, could be improved by some clarifications:

1. the authors keep talking about 'large scale systems' and Grids.
However, these large-scale systems (and their relation to Grids!) are never really explained/defined/motivated.

2. Having 20 responses is a lot for a user survey. But it is by far too little to do statistics with it. (1 respondant 
already contributes 5% -- so terms like '62% of all' does absolutely not make sense and should be avoided for scientific
 soundness)

I would strongly recommend to rephrase all sentences mentioning percentages in favour of more fuzzy terms. There are 
some strong trends to report here, but these shold not be put into doubt by presenting less clear results with the same 
level of (fake) confidentiality.

3. Also, I would recommend merging Sections 3 and 4. Currently, the reader has to keep jumping back and forth. I would 
propose to start all results with the respective graph from Section 4, followed by the interpretation from Section 3.

4. Finally, there are some terms that need to be explained. E.g., neither UML nor HTML are tools, but they are mentioned
 as design tools. Did the people use UML-based tools? or UML directly??

Under 'maintenance and administration cycle' PBS is listed as a tool. How can one use a scheduler for maintenance ??