OK, but initially trying to benchmark on a t1.micro indicates a lack of experience with EC2 that colors any other benchmarking you might be doing.

Add in others' comments about this basically compromising security for speed and this is a bit of an irresponsible article.