A Performance Evaluation of Alternative Mapping Schemes for Storing XML Data in a Relational Database

TLDR

XML is becoming a dominant data format for internet processing, and query languages such as XQL, Lorel, XML‑QL, and XML‑GL have been proposed to query XML data. The study investigates how XML data can be stored and queried using a standard relational database system. The authors present alternative mapping schemes for storing XML in a relational database, explain how XML‑QL queries translate to SQL for each scheme, and conduct performance experiments evaluating database size, query performance, and update performance. The experiments show that a variant of a binary mapping scheme delivers the best overall performance, and the results apply broadly to semi‑structured data models and query languages.

Abstract

XML is emerging as one of the dominant data formats for data processing on the Internet. To query XML data, query languages like XQL, Lorel, XML-QL, or XML-GL have been proposed. In this paper, we study how XML data can be stored and queried using a standard relational database system. For this purpose, we present alternative mapping schemes to store XML data in a relational database and discuss how XML-QL queries can be translated into SQL queries for every mapping scheme. We present the results of comprehen- sive performance experiments that analyze the tradeoffs of the alternative mapping schemes in terms of database size, query performance and update performance. % The results show clearly that a variant of a binary mapping scheme % shows the overall best performance While our discussion is focussed on XML and XML-QL, the results of this paper are relevant for most semi-struct- ured data models and most query languages for semi-structured data.