Bulkloading and Maintaining XML Documents

The popularity of XML as a exchange and storage format brings about massive amounts of documents to be stored, maintained and analyzed -- a challenge that traditionally has been tackled with Database Management Systems (DBMS). To open up the content of XML documents to analysis with declarative query languages, efficient bulk loading techniques are necessary. Database technology has traditionally been offering support for these tasks but yet falls short of providing efficient automation techniques for the challenges that large collections of XML data raise. As storage back-end, many applications rely on relational databases, which are designed towards large data volumes. This paper studies the bulk load and update algorithms for XML data stored in relational format and outlines opportunities and problems. We investigate both (1) bulk insertion and deletion as well as (2) updates in the form of edit scripts which heavily use pointer-chasing techniques which often are considered orthogonal to the algebraic operations relational databases are optimized for. To get the most out of relational database systems, we show that one should make careful use of edit scripts and replace them with bulk operations if more than a very small portion of the database is updated. We implemented our ideas on top of the Monet Database System and benchmarked their performance.