An Exploratory-Descriptive Review of Main Big Data Analytics Reference Architectures – an IT Service Management Approach

Big Data Analytics (BDA) aims to create decision-making business value by applying multiple analytical procedures from the Statistics, Operations Research and Artificial Intelligence disciplines to huge internal and external business datasets. However, BDA requires high investments in IT resources – computing, storage, network, software, data, and environment -, and consequently the selection of the right-sized implementation is a hard business managerial decision. Parallelly, IT Service Management (ITSM) frameworks have provided best processes-practices to deliver value to end-users through the concept of IT services, and the provision of BDA as Service (BDAaaS) has now emerged. Consequently, from a dual BDA-ITSM perspective, delivering BDAaaS demands the design and implementation of a concrete BDAaaS architecture. Practitioner and academic literature on BDAaaS architectures is abundant but fragmented, disperse and uses a non-standard terminology. ITSM managers and academics involved on the problematic to deliver BDAaaS, thus, face the lack of mature practical guidelines and theoretical frameworks on BDAaaS architectures. In this research, consequently, with an exploratory-descriptive purpose, we contributed with an updated review of three main non-proprietary BDAaaS reference architectures to ITSM managers, and with a hybrid functional-deployment architectural view to the BDAaaS literature. However, given its exploratory status, further conceptual and empirical research is encouraged