1. Introduction
The main characteristics of big data described through the “four V's of volume, variety, velocity and veracity” [1] have steered the discussion and the development of big data techniques into big computing infrastructure (e.g., high performance and data intensive computing/cloud systems), big data storage and scalable data structures (e.g., BigTable and Cassandra), scalable computation frameworks (e.g., Hadoop/MapReduce and S4), and scalable data mining algorithms [2]–[6]. However, few discussions have been focused on dynamic and flexible data analytics processes that rely on multi-dimensional elasticity perspectives from consumers and providers, while leveraging these existing powerful computing infrastructures, frameworks, and algorithms. We argue that elasticity principles, such as, resource, quality and cost elasticity [7], should be investigated as fundamental guidelines for developing new data analytics platforms to tackle issues in big data analytics, for example: