Self-organized and fully service-automated monitoring approach at the cloud-network slice granularity

Abstract

The cloud-network slicing concept has been established to deal with the advent of the Fifth Generation (5G) of mobile networks and its enabling technologies, thus promoting softwarization and cloudification. The Novel Enablers for Cloud Slicing (NECOS) ecosystem distinguishes itself over state of the art through the definition of slicing at both cloud and network levels and by promoting a Management and Orchestration (MANO) platform that provisions features with a self-organized and full-service automation approach across multiple federated domains. In the NECOS architecture, the Infrastructure and Monitoring Abstraction (IMA) component fetches Key Performance Indicators (KPIs) associated with the constituent parts of the active cloud-network slice instances managed by the NECOS platform. However, the design of the IMA monitoring component follows a centralized approach running at the core-cloud domain. Thus, the IMA concentrates on the monitoring data fetching function, done through interaction with measurement applications. It books all of them into a database and then forwards the incoming data for targeting management applications. Our findings in the slice monitoring state of the art study and assessments in IMA central cloud monitoring suggest that the centralized cloud approach cannot make distinct monitoring technologies compatible, besides not presenting a monitoring-as-a-service perspective that allows a self-organized and fully-service automated monitoring management scheme. In this regard, this work proposes the Distributed Infrastructure and Monitoring Abstraction (DIMA) multilevel monitoring plan. DIMA can promote monitoring as a service across an edge-cloud continuum inside the NECOS domain, enabling cloud/edge-centric and distributed monitoring schemes at the granularity of cloud-network slices' constituent parts.