I'm moving this discussion to the thread as suggested by the Lucene mentors.
This is my final proposal link: http://www.google-melange.com/gsoc/proposal/review/google/gsoc2011/varunthacker1989/1 Till now I have implemented the following to get a better understanding of my project: I wrote a sample code to test out speed difference between SEQUENTIAL and O_DIRECT( I used the madvise flag-MADV_DONTNEED) reads .This is the link to the code: http://pastebin.com/8QywKGyS. There was a noticeable speed difference when i switched between the two flags. I did not use the O_DIRECT flag because Linus Torvalds had criticized it. This blog post by Micheal McCandless ( http://blog.mikemccandless.com/2010/06/lucene-and-fadvisemadvise.html) helped me understand the problem for which LUCENE-2793 is needed. I am currently reading up on the Lucene Documentation taking into account a few pointers provided by Simon WIllnauer. I want to keep everyone updated and also take into consideration the comments made by the members of this community, which will help me understand and implement these tasks . -- Regards, Varun Thacker http://varunthacker.wordpress.com