One thing that I would find interesting would be comparing different techniques for parallel programming. e.g. MapReduce vs. JavaSpace vs. GPU vs. SGE vs Others
The paper could talk about what type of services work best for what type of problem (Data intensive Vs. Computational Intensive). Comparing and contrasting setup, ease of scalability, return on investment for adding additional resources (Linear, non-linear). As a final part you could talk about possibly comparing two or more of the techniques together. As someone who works on both computationally intense and data intensive problems. I would find this paper a good resource, and others probably would as well. And if nothing else you would get exposure to several different technologies which always looks good on the CV. Just my $0.02 Jeremy R. Easton-Marks "être fort pour être utile" On Sat, Oct 15, 2011 at 8:27 AM, Zsolt Kúti <[email protected]> wrote: > Although the orginal poster seeks for concrete problems for suggestion, it > is worth mentioning here a great material that gives a foundation for the > space approach. It provides an exhaustive description, far more that could > be done in a forum. > > How to write parallel programs - A first course > http://www.lindaspaces.com/book/ > <http://www.lindaspaces.com/book/> > Zsolt >
