Abstrakt
Huge amount of data is generated by current IT technologies and systems, mainly in a form of various logs or user data. Processing, storing and analysis of this data became problematic using traditional database systems.
Bobox is a parallel framework designed to support development of data-intensive parallel computations. The main idea behind Bobox is to connect a large number of relatively simple computational components into a nonlinear pipeline and communicating using messages.
This schema may be applied also in a distributed environment. The paper presents requirements on the Bobox and its extension called D-Bobox based on an analysis of extensibility and potential problems of the extension with automatic plan generation in a distributed environment kept in mind.