|All three previous editions had an introductory
chapter called "The Roots" that included some of the field's most
seminal papers. How could you eliminate that material from this
This was a difficult decision for two reasons. First, the well-educated database researcher should have an intimate familiarity with this work because it shaped the field and continues to frame much of the discussion. Second, it is educational for researchers today to read these original papers and attempt to enter the mindset of the researchers working on the early systems, especially to see how they wrestled with the as-yet unformed structure of the design space. (The INGRES papers are especially nice in that regard.) Unfortunately, I found that teaching these papers -- especially as the first papers in the class -- was a turnoff for many of today's students. It is fairly difficult to separate the wheat from the chaff in these papers on first reading, and they do not give students a good sense of the structure of the field that emerged from this work. Our new introductory chapter attempts to better set the stage of the field, and put the remaining papers into context. We encourage serious database experts to go read the seminal papers as well.
|I have heard that the Berkeley graduate
curriculum no longer follows this book. What is the story
Traditionally, Berkeley had graduate-level reading/project courses in both Operating Systems (CS262) and Databases (CS286). Through editions 1-3, Readings in Database Systems reflected the CS286 database course at Berkeley.
During that period, a large fraction of the students took graduate OS, and a more select group took Databases. In 1999, Eric Brewer and I tried an experiment where we merged the curricula of these two courses into a single course offering, spread over two semesters. The idea was to provide students (and the instructors!) a more holistic view of software systems by looking at core issues through the lens of two different communities. We alternated giving lectures, attending each other's lectures and heckling from the audience to highlight the differences in worldview between OS and DB researchers. This role-playing stimulated good student discussion and shed light on a larger computer systems design space than is traditionally considered by either community.
The newly retooled course was very well received by students, and there have been significant tangible results in the database and systems research coming out of Berkeley in the last five years. It has also changed my research thought processes in a positive way. I recommend the joint course highly, especially when coupled with advanced topics courses in databases.
In recent years, we have only offered the first semester of the course jointly. We are leaning toward reinstituting an advance CS286 to cover the remaining DB material in more depth during the second semester.
A recent web page for the course is up at Eric Brewer's website.
introductions were written by Hellerstein and which by
This is left as an elementary exercise for the reader.
© 2005, Joseph M. Hellerstein.