Highly available database system configurations, which include active and standby systems, are often deployed in environments in which the active and standby systems have access to shared persistent storage. However, the widely used log-shipping approach to high availability does not exploit such shared storage. We propose a high availability technique for database systems that allows the active and standby systems to collectively manage a single shared copy of the database. This approach simplifies database high availability by replacing two instances of the database with one. Furthermore, it has the potential to outperform log shipping by allowing the active and standby systems to share burden of updating the database.