GB979D Big Data Analytics Big Data Repository R16.5.1

The Analytics Big Data Repository (ABDR) is a CSP’s data repository, used mainly for analytical purposes, permitting efficient and straight forward re-use of data for multiple purposes. The ABDR is a new concept being developed and specified by the TM Forum in order to permit the creation of standardized implementations across operators, and straightforward re-usability of the data by ABDR compliant solutions.

Informally, an ABDR implementation is a collection of unique, multiple, independent data entities that have a clear definition, i.e. a data dictionary, per data entity. For example an ABDR can include CDRs, DPI, and billing records with a data dictionary defining each of these record types. The data dictionaries and data entities are not arbitrary but are established according to the ABDR definition being developed by the TM Forum and its members.

Many systems have been designed over time in order to deal with analytics of large sets of data, each with their own objectives and operating environments. Big data analytics (BDA) represents a specific — and novel — set of goals and challenges that lead to particular approaches to projects, data organization and platform criteria.

This Addendum will develop the notion of an ABDR and explore the specificity of big data analytics, especially with respect to other analytics technologies and approaches such as those related to Enterprise Data Warehouse or Hadoop type platforms. The term ABDR will be used to encompass both organizational and technological related topics since they are intertwined.

The Addendum starts with a review of the successive generations of platforms and highlights trade-offs that ABDR architects and users face. The TM Forum has conducted extensive research in the area of Big Data Analytics, and this document, GB979D, builds on that research by illustrating how the main challenges of big data analytics imply new approaches that are emerging as the maturity in this domain develops.

The Addendum will conclude with some considerations on the current trends and perspectives for big data analytics, as this domain is still in an early stage of development.