The AutoAdmin research project was launched in the Fall of 1996 in Microsoft Research with the goal of making database systems significantly more self-tuning. Initially, we focused on automating the physical design for relational databases. Our research effort led to successful incorporation of our tuning technology in Microsoft SQL Server and was subsequently also followed by similar functionality in other relational DBMS products. In 1998, we developed the concept of self-tuning histograms, which remains an active research topic. We also attempted to deepen our understanding of monitoring infrastructure in the relational DBMS context as this is one of the core foundations of the “monitor-diagnose-tune” paradigm needed for making relational DBMS self tuning. This article gives an overview of our progress in the above three areas – physical database design, self-tuning histograms and monitoring infrastructure.