In the past six years or so, I've moved from academia to industrial research and now to an Internet startup. In this talk, I'll share some of the experiences I've had and some of the things that I've observed and learned along the way. Among the issues to be discussed are some of the technical and nontechnical "real-world constraints" that I've bumped into in trying to apply research-thinking in practice. I'll also share some thoughts on where database systems may or may not be headed as a result. I'll close by describing some of what we're doing in "database space" at Propel, including how we're using database systems as off-the-shelf parts in assembling what we like to think of as "the e-commerce platform of the future".
This talk is a lightly updated "re-run" of my VLDB'00 talk (for those who weren't in Egypt last fall).
Mike got his PhD from UC Berkeley in 1983. He spent 1983-95 on the UW-Madison Computer Sciences Department faculty (with a sabbatical year at the IBM Almaden Research Center from 1993-94); there he worked with students and colleagues on a variety of database system issues, mostly related in some way to either database system performance or object databases. While there, he helped clear several forests by publishing numerous papers in ACM SIGMOD, VLDB, and other such venues. Mike then spent the five years from 1995-2000 at IBM Almaden working on object-relational databases; he helped to ensure that neither SQL99 or DB2 UDB would end up having too few features. In 2000 Mike joined an internet infrastructure startup, Propel (www.propel.com), just in time to watch the internet economy crash from the inside rather than the outside.