A Scalable Telemetry Framework for Zero Touch Optical Network Management

Abstract

The interest about Zero Touch Network and Service Management (ZSM) is rapidly emerging. As defined by ETSI, the ZSM architecture is based on a closed-loop/feedback control of the network and the services. Such closed-loop control can be based on the Boyd's Observe Orient Decide and Act (OODA) loop that matches some specific management functions such as Data Collection, Data Analytics, Intelligence, Orchestration and Control. An efficient implementation of such control loop allows the network to timely adapt to changes and maintain the required quality of service.Many solutions for collecting network parameters (i.e., implementing ZSM data collection) are proposed that fall under the broad umbrella of network telemetry. An example is the Google gRPC, that represented one of the first solutions to provide a framework for data collection. Since then, the number of available frameworks is proliferating. In this paper we propose the utilisation of Apache Kafka as a framework for collecting optical network parameters. Then, the paper goes beyond that by proposing and showing how Apache Kafka can be effective for supporting data exchange and management of whole ZSM closed-loop.Experimental evaluation results show that, even when a large number of data are collected, the solution is scalable and the time to disseminate the parameter values is short. Indeed, the difference between the reception time and the generation time of data is, on average, 40-50ms when about four thousand messages are generated.

360° View on Zero-Touch (Zero-Touch) Networks

“…The ZSM group was introduced by ETSI in December 2017 to speed up the definition of the necessary architectures and solutions. Zero-touch networks and service management are gaining popularity quickly, as per research [5].…”

A review of machine learning-based failure management in optical networks

“…Thus, the performance of learning algorithms depends heavily on data quantity and quality. In optical networks, data can come in several forms that can be roughly divided into numerical data, time series data, text data, image data, and statistical data [18,19]. The realization of failure management is based on the collected data pertaining to all elements and over the entire lifetime of the system.…”

Maximizing Network Capacity, Control and Management in Designing a Telemedicine Network: A Review and Recent Challenges

“…This type of network combines the power of optical network technologies and cloud computing. This type of network is utilized to provide telemedicine applications with high-speed local networks [24,25]. For high-speed, high-quality, and high-capacity internet traffic, [26] proposes an impairmentaware reactive defragmentation strategy for dynamic elastic optical networks servicing telehealth care-oriented traffic.…”