The BigDawg Architecture and Reference Implementation

This paper presents the reference implementation of a new architecture for future “Big Data” applications. Such applications require “big analytics” as one might expect, but they also require real-time streaming support, real-time analytics, data visualization, and cross-storage queries. We are guided by the principle “one size does not fit all”, and we build on top of three storage engines, each designed for specialized use cases. In addition, we demonstrate novel support for querying across multiple storage engines as well as pioneering solutions to data visualization. In the remainder of this short paper, we describe the first of three BigDawg reference implementations, Bulldog. In the next two years we expect to follow with Pitbull and Rottweiler releases.