Paper Info
Title
Building a Native XML-DBMS as a Term Project in a Database Systems Course
Abstract
This is to report on a database systems course the rst author held in the summer semester of 2005 at Saarland University, Saarbr?ucken, Germany. This course was an experiment in several respects. For one, we wanted to teach a systems course with a practical part in which students apply the material taught to build the core of a database management system. Such a systems building e ort seems to be quite common in top-tier US uni-versities, but it is rare in Europe. One main reason for this is that European curricula often require students to take many small courses per term. Students then cannot be required to invest the time necessary for such a systems-building e ort into an individual course. In Saarbr?ucken, this fortunately does not apply and students are expected to take only about two main courses per term. (The database systems course in Saarbr?ucken is worth 9 points in the European course-credit transfer system ECTS, which corresponds to an estimated workload of 20 hours per week. ) Second we wanted our students to do something reason-ably new to facilitate follow-up bachelor's and master's pro-jects. So we decided to have them build the core of a native XML-DBMS, an active research area with a number of ex-citing problems. This would allow students to do something that may never have been done before: Indeed, to our knowl-edge, no database systems course has had students build a native XML-DBMS before. (However, we are aware of a compiler construction course at UC San Diego in which a main-memory XQuery system was built [7]. ) A main goal of the course was to convince the participants that systems research goes far beyond good programming. Students should get the opportunity to experience success in speeding up query evaluation by several orders of magnitude by using the techniques and algorithms taught in the course. Even though this course was meant to be an intensive ex-perience, we had to make compromises to make the project feasible. We decided to restrict the project to building the query processor, but keep updates as simple as possible and completely disregard concurrency control and recovery. We describe our design decisions and our ways in which we have shaped the project in the remainder of this paper.
Year
Venue
Keywords
2006
XIME-P
database management system,compiler construction,database system,concurrency control
Field
DocType
Citations 
Database tuning,Database model,XML,Computer science,View,Component-oriented database,Database design,Database schema,Database theory,Database
Conference
0
PageRank 
References 
Authors
0.34
3
3
Name
Order
Citations
PageRank
Christoph Koch12817163.43
Dan Olteanu2129974.52
Stefanie Scherzinger320920.82