A workbench for advanced database implementation and benchmarking

This work focuses on a methodology to help bring many of our database artifacts and prototypes to reside on the top of a common workbench platform that leads to uniformity and removes overlap across different subsystems. A versatile command format has been developed to allow commands belonging to different subsystems to be interleaved in the same batch unambiguously. Through a collaborative effort carried on in parallel, existing GUIs (Graphical User Interfaces) have also been merged into a common, but simple GUI. The GUI executes a batch of commands. Subsystem currently included are: a runner for SQL on a variety of database platforms, a runner for Quilt queries (Quilt is an early version of XQuery and runs on a platform called KWEELT), ElementalDB, an experimental database system used for instruction in a graduate database implementation course, our own XQuery engine which aims at handling data in terabyte range stored in our storage in paginated form using our pagination algorithm, a research prototype for NC94, an important spatiotemporal data set in agriculture, and a research prototype for a temporal database. The organization of the subsystems follows strict convention for ease of further development and maintenance. XML is used extensively by various subsystems. An XML based framework has been developed for benchmarking subsystems to make experiments completely repeatable at click of a button starting from creation of storage, loading of data sets, execution of commands, collecting performance data in XML-based logs to reporting using XQuery queries on the XML logs. With a very small learning curve, the resulting workbench can be used by students, instructors, developers and researchers alike and managed easily