In this paper, we present our work on evaluating the skyline algorithms BNL, SFS, and a variant of LESS in PostgreSQL. It is well known that the performance of skyline queries is sensitive to a number of parameters. From extensive experiments on skyline implementations we have discovered several rules, which are remarkably simple and useful, but hard to obtain from theoretical investigation. Our findings are beneficial for developing heuristics for the skyline query optimization, and in the meantime, provide some insight for a deeper understanding of the skyline query characteristics.