English abstract
XML data stored in huge containers is classified as Big-Data. This data can appear in structured, semi-structured, and unstructured format so they cannot be processed in traditional relational database management systems. However, they can be processed using Big-Data management Systems such as Marklogic.XML features such as breadth and depth as well as XQuery categories can have an impact on Big-Data management systems. This research aims into identifying the features that can have an impact on these systems. To achieve this objective Marklogic, a Big-Data management system, is used to execute certain classes of XML queries over different categories of XML databases. The performance of Marklogic is measured by recording the queries elapsed time. Both datasets and query sets are taken from Extended 3D~XBech benchmark which has proven its superiority among the existing XML benchmarks.
The experimental results have shown that the breadth of the underlying XML database is a bottle-neck for the tested query types. In addition, the performance of the "Join-on-Values" query was too low, whereas the location of the sought-data in the database does not affect the performance much.As a future work, the experiment can be extended by testing more XQuery types and/or considering more metrics such as CPU usage, memory consumption, and IO-Operations. In addition, the experiment can be tested in a single or distributed system. Finally, further testing of different Big-Data NoSQL databases will help to give more confident on the highlighted results.
