The article discusses the streaming data processing in large quantities for the situation-oriented databases (SODB). Traditionally SODB provides cached processing of heterogeneous data. This article considered the work with large files that do not fit entirely in memory – containing a number of similar fragments, which can be processed portions. Portions of data are extracted from the input stream, processed in the buffer, and then sent to the output stream. Three variants of implementing this scheme are considered in the framework of the hierarchical situational model of the HSM. The tools offered at the conceptual level are illustrated by the example of processing XML data using PHP software tools, such as XMLReader, XMLWriter, DOM.