Perfomance of XSLT processors on large data sets

An XML transformation, commonly specified in XSLT, is an important step during XML-based data migration. Since common XSLT processors always parse input data into DOM-like structure, processing of large data sets to be migrated represents a significant problem; all available memory can be exhausted, the trans-formation takes unacceptable long time, or even the processors can fail. In this survey we explore current state-of-the-art in XML transformation processing on large data sets and present a performance comparison of DOM-based XSLT processors as well as streaming XSLT and STX processors. We show that there is no silver bullet among XSLT processing on large data sets; particular approaches have their strengths and limitations.