In this paper we discuss advances in self-tuning database systems over the past decade, based on our experience in the AutoAdmin project at Microsoft Research. This paper primarily focuses on the problem of automated physical database design. We also highlight
other areas where research on self-tuning database technology has made significant progress. We conclude with our thoughts on opportunities and open issues.