Optimizing queries on files

Abstract

We present a framework which allows the user to access and manipulate data uniformly, regardless of whether it resides in a database or in the file system (or in both). A key issue is the performance of the system. We show that text indexing, combined with newly developed optimization techniques, can be used to provide an efficient high level interface to information stored in files. Furthermore, using these techniques, some queries can be evaluated significantly faster than in standard database implementations. We also study the tradeoff between efficiency and the amount of indexing.