Abstract: A network of data processing nodes is configured for in-memory data processing. The data processing nodes are interconnected in an acyclic manner, and are connected at an input end to one or more input data sources, and at an output end to one or more output data nodes. Each data processing node is configured to perform one or more data processing operations which generate respective result data that is used as one or both of (i) an input for another data processing node in the network, and (ii) an output to be sent to one of the output data nodes. The result data for each data processing node is stored in a memory of the respective data processing node. Each data processing node includes a flag having a state that controls storage persistence of the result data in the memory of the respective data processing node.

Abstract: A method and computer program product are provided for defining a dynamic data storage class in-memory using a strongly-typed compiled language, and accessing data in the memory. A schema is defined of one or more key-value pairs of field names and data types, wherein the schema represents columns of a table in the memory. A table key is declared that represents the table in the memory. The dynamic data storage class is instantiated, using the strongly-typed compiled language, by passing the table key. Each instantiation of the dynamic data storage class is an individual row of the table in the memory. Fields of the dynamic data storage class are accessed by an overloaded indexer that uses strings as the field names in the schema.