Leveraging Industry Benchmarks to Teach Database Concepts

Databases have been a part of core curriculum in computer science for decades. Courses exist at both introductory and advanced levels and topics include data modeling and design, query optimization, and studying features of specific database platforms such as MySQL. Concepts such as normalization in relational databases, OLTP/OLAP are taught using assignments that are based on short synthetic examples. In this talk, we propose the use of open use industry benchmark kits (HammerDB), the Wisconsin benchmark, and/or TPC-defined industry benchmark to create "close to real-world" and engaging course modules. The use of benchmarks will not only offer new ways of teaching the existing topics but also allow extending the scope of current database courses to contemporary data management systems and big data technologies. Students will be exposed to a breadth of areas including database internals, performance optimization, schema design (real world schemas with "large" datasets), an insight into the underlying system and its performance, and exposure to -what's possible" with current hardware all under one course. The three benchmarks we discuss cover three different data management scenarios. TPC-[C/E] serves OLTP use-case, TPC-[H/DS] addresses OLAP while TPCx-BB supports a mix of OLAP and Big Data. We experimented with this idea by organizing two guest lectures on database benchmarks to a class of computer science students at Portland State University; and it was very well-received as evident from students' feedback. Through this proposal, we hope to invite interested faculty for further discussions so that it may develop into a multi-institutional endeavor.