Implementing Filesystems by Tree-aware DBMSs

Abstract

With the rise of XML, the database community has been challenged by semi-structured data processing. Since the data type behind XML is the tree, state-of-the-art RDBMSs have learned to deal with such data (e.g., [18, 5, 6, 16]). This paper introduces a Ph.D. project focused on the question in how far the tree-awareness of recent RDBMSs can be used to, once again, try to implement filesystems using database technology. Our main goal is to provide means to query the data stored in filesystems and to find ways to enhance/ combine the data storage and query capabilities of operating systems using semi-structured database technology. Two DBMSs with relational XML storage, built on top of the XPath accelerator numbering scheme [14], are the foundations for our work. With BaseX, an XML database, we establish a link between user, database and lesystem content. BaseX allows visual access to filesystem data stored in the database. An integrated query interface allows users to filter query results in interactive response time. Second, we establish a link between DBMS and OS. We implement a filesystem in userspace backed by the MonetDB/XQuery system, a well-known relational database system, which integrates the Pathfinder XQuery compiler [5] and the MonetDB kernel [4]. As a result, the DBMS is mounted as a conventional filesystem by the operating system kernel. Consequently, access via the established (virtual) filesystem interface as well as database enhanced access to the same data is provided.