Growing number of applications store data in relational databases. Moving database applications to the cloud faces challenges related to flexible and scalable management of data. The obvious strategy of hosting legacy database management systems (DMBSs) on virtualized cloud resources leads to sub optimal utilization and performance. However, the layered architecture inside the DBMS allows for virtualization and consolidation above the OS level which can lead to significantly better system utilization and application performance. Finding an optimal database cloud solution requires finding an assignment from virtual to physical resources as well as configurations for all components. Our goal is to provide a virtualization advisor that aids in setting up and operating a database cloud. By formulating analytic cost, workload, and resource models performance of cloud-hosted relational database services can be significantly improved.