Big Data over Named Data Networking

5 février 2018

Junior Dongo (LACL)

Information Centric Networking (ICN) aims to be the next Internet architecture. It brings many features such as network scalability and in-network caching by moving from the networking paradigm from the current host centric to content centric where the resources are important not their location. In order to benefit from the ICN properties, Big Data architecture needs to be adapted to comply with this new Internet architecture.

In this talk, our current effort to define Big Data architecture operating on Named Data Networking (NDN) and capitalizing on its properties. We present a fully distributed, resilient, secure and adaptable distributed file system NDFS (NDN Distributed File System) which is the first layer (Data Layer) in the Big Data stack. To perform computation on the data replicated using a Distributed File System, a Big Data architecture must include a Compute Layer. Based on NDN, N-MapReduce is a new way to process large datasets of information. It has been designed to leverage the features of the data layer.