|
Minibase History
Many years ago, David DeWitt developed a database course project that involved building a small relational DBMS called Minirel in C. It has inspired several revised versions, and these have been used at Wisconsin and elsewhere over the years. Mike Carey and Raghu Ramakrishnan wanted to develop a more complete version of Minirel, in C++, including components for concurrency control and recovery, more relational operator evaluation methods, a query optimizer, and design tools. The result is Minibase. Minirel supported a course project in which students developed parts of the code; Minibase additionally supports exercises in which various components are studied through experiments. The system design differs in many ways from earlier versions, and it is a more comprehensive system than Minirel, but hopefully, the spirit of the original design, encouraging a hands-on approach to databases, is retained. Much of the new Minibase was developed as a graduate course project by students taking CS764. Different groups developed different parts such as the disk space and heap file managers (Ravi Murthy and Sriram Narasimhan), the buffer manager (Kan Qiu and Fei Su), B+ trees (Yanming Cao and Celena Tanguay), linear hashing (Weiqing Huang), log manager (Cjin Lee and Ajitkumar Natarajan), lock manager (Francis Salmon and Gary Zipfel), recovery manager (Ranjani Ramamurthy and Arthur Tsoi), join methods (Raghu Mallena and Amit Shukla) and query plan interpreter (Donko Donjerkovic). Peter Feakins and Bill Kimmel were responsible for integrating the various components for the single user minibase. The project was coordinated by Mike Carey and Raghu Ramakrishnan. Some code (e.g., parts of the buffer manager and heap files) was based on a C++ version of Minirel developed by DeWitt and Jussi Myllymaki, but most of it was written afresh. The code developed in CS764 was significantly refined and extended, and several additional tools were designed and developed, through the efforts of many people under the direction of Ramakrishnan. Huseyin Bektas and Harry Stavropoulos developed visualization tools for the buffer manager and B+ trees, and this code was developed further by Weiqing Huang. Stephen Harris and Michael Lee developed an SQL query optimizer and a visual plan explanation tool. Donko Donjerkovic developed the query optimizer further. Bill Kimmel and Luke Blanshard continued the integration effort and considerably cleaned up the single-user version. Ranjani Ramamurthy substantially reworked many of the multi-user components and integrated them to produce the multi-user version of Minibase. Eben Haber used the Opossum schema manager to develop a tool for drawing and maintaining ER diagrams. Andy Therber and Andrew Prock developed an ER diagram to relational tables translator and a Tcl/Tk based database design tool that supports normalization. Also, Emmanuel Ackaouy, Lee Schumacher and Bryan So made significant contributions to an earlier version of Minibase. This code from this version did not survive, but it influenced the design of the current Minibase. Venkatesh Ganti, Jonathan Goldstein, Krishna Kunchithapadam and Uri Shaft contributed to the testing and development of this software. Shaun Flisakowski, in particular, significantly revised and re-organized Version 2.0 of Minibase. Needless to say, Minibase would not have been possible without the dedicated efforts of the many students involved. Raghu Ramakrishnan [raghu@cs.wisc.edu]
Back to the Minibase Home Page
|