A Survey on XML Fragmentation

Abstract

Efficient document processing is a must when large volumes of XML data are involved. In such critical scenarios, a well-known solution to this problem is to distribute (map) the data among several processing nodes, and then distribute the processing accordingly, taking advantage of parallelism. This is the approach taken by distributed databases and MapReduce environments. Fragmentation techniques play an important role in these scenarios. They provide a way to "cut" the database into pieces and distribute the pieces over a network. This way, queries can also be "cut" into sub-queries that run in parallel, thus achieving better performance when compared to the centralized environment. However, there is no consensus in the database community as to what an XML fragment is. In fact, several approaches in literature present definitions of XML fragments. In addition to query processing, using XML fragmentation techniques may also be helpful when managing XML documents distributed along the web or clouds. This paper surveys the existing XML fragmentation approaches in literature, comparing their features and highlighting their drawbacks. Our contribution resides in establishing a map of the area.