Minimizing Time-Space Cost For Database Version Control

Lin Yu, Daniel J. Rosenkrantz

Abstract: We introduce the concept of a version graph to model the problem of minimising the space and version regeneration cost for database version control. We show that, in general, this problem and several of its variations are NP-complete. Motivated by the practical importance of these problems, we develop several heuristics and obtain worst-case guarantees on their performance. We also present linear time algorithms for problems characterized by special classes of version graphs.