Automata-based dynamic data processing for clouds

Abstract

Big Data is a challenge in many dimensions one of which is its processing. The often complicated and lengthy processing requires specialized programming paradigms to distribute and scale the computing power. Such paradigms often focus on ordering tasks to fit an underlying architecture. In this paper, we propose a new and complementary way of reasoning about data processing by describing complex data processing as set of state transitions, specifically, a non-deterministic finite automata which captures the essence of complex data processing. Through a P2P implementation of this model we demonstrate the dynamism, intrinsic scalability and adeptness to Cloud architectures.