CANDIL: A federated data fabric for network analytics

Abstract

The availability of data sources during the Big Data era provides the opportunity for new analytical applications in the networking domain, which are envisioned as one of the main enablers of the future autonomous networks. But the proliferation of heterogeneous data sources has resulted into a sea of data silos, in which finding data, understanding data, and dealing with the complexities of each data source becomes a challenge. Aiming to tackle the connection of data silos, the data fabric has appeared as a new paradigm that provides a uniform access to all the data, abstracting consumers from the underlying complexities of the data sources. In this regard, the knowledge graph has raised as a promising solution that can integrate data from heterogeneous silos based on common concepts captured in ontologies. Building upon knowledge graph standards, this paper introduces CANDIL, a federated data fabric to support the integration of data from distributed systems, mainly focused on networking domain aspects. CANDIL defines an ontology that captures network topology and interface concepts, along with a reference architecture to ingest and integrate data in a federated knowledge graph that spans across the Edge-Cloud continuum. The proposal is validated with a prototype implementation and two example use cases of network analytics.