A goal-oriented self-tuning caching algorithm

The contribution of this paper is a novel approach to adaptivity that combines alternatives rather than selecting one among alternatives. Using only three, homogenous, cache replacement algorithms, GD-GhOST were able to provide a cache replacement policy that requires no tuning or user-intervention beyond the initial selection of the performance criteria to be optimized. Overall, at its worst observed performance GD-GhOST was within approximately 1% of the best policy´s miss ratio, and at its best, GD-GhOST reduced byte miss rates by well over 50%.