In data-intensive industrial on-line applications utilizing live process data, one faces an unusual set of database requirements. The process measurement data need to be acquired at great speed, organized in time series and made available for time-based retrieval. Active capabilities and functional extensibility are needed to implement a flexible data-driven processing paradigm. An efficient transaction logging and recovery mechanism is needed in order not to impede the data acquisition flow. RapidBase is a system that meets these requirements. It utilizes a main-memory database, a unique temporal-relational data model for handling time series, and an elaborate trigger subsystem. It is implemented as a server program equipped with interfaces of high power of expression.