In this section, we describe how an in-disk scheduler must evolve to support range writes. We present two approaches. The first we call expand-and-cancel scheduling, a technique that is simple, integrates well with existing schedulers, and performs well, but may be too computationally intensive. Because of this weakness, we present a competing approach known as hierarchical-range scheduling, which requires a more extensive restructuring of a scheduler to become range aware but thus avoids excessive computational overheads.
Note that we focus on obtaining the highest performance possible, and thus consider variants of shortest-positioning-time-first schedulers (SPTF). Standard modifications could be made to address fairness issues [18,28].