Abstract
The processing of growing volumes of technological data in industrial automation systems, as well as the limitations of traditional archiving in databases with deferred analysis, are considered, taking into account time requirements and infrastructure load. The concepts of stream processing are generalized and adapted to technological measurements and events in automated control systems. A multi-level model is presented that combines data source levels, SCADA, and storage subsystems. A test bench demonstrates a reduction in the intensity of writing to a relational database while preserving characteristic signal changes and similar statistical delay indicators. Tabl.: 2. Refs.: 23 items.