Towards a User-Friendly Loading System for the Analysis of Big Data in the Internet of Things

Abstract

In the Internet of Things (IoT) era, we need to face increased masses of cross-domain data stored in different formats (either relational, XML, JSON, textual) and data streams (produced by sensors), that can be highly or loosely structured and need to be integrated for analysis. Recently, many NoSQL systems (e.g., MongoDB, Cassandra, HBASE) have been born for coping the scalability issues of current analysis approaches. They all are based on different data models and present different performances based on structure, size and location of data distributed on clusters. While the research community is really active in the development of techniques for big data analysis, at current stage a comprehensive solution that supports the user in loading data from heterogeneous sources and in integrating them into the most suitable NoSQL system is still lacking. In this paper we propose the design and expected functionalities of Big Loader, a user-friendly loading system for NoSQL systems that allows the specification of the conceptual schema of the data to be loaded, the specification of the sources from which the data should be gathered in order to feed the conceptual schema and finally we outline intelligent strategies for the selection of the NoSQL system where the conceptual scheme can be deployed.