Network Telemetry Streaming Services in SDN-Based Disaggregated Optical Networks

Abstract

Accurate real-time availability of transmission parameters at the network controller has the potential to significantly improve the efficiency of control and management operations, particularly in the case of soft failures where detection and localization procedures are typically affected by the long monitoring time intervals usually adopted in current Elastic Optical Networks (EONs). In this paper, we report on the design, implementation and experimental demonstration of a telemetry service exploiting the gRPC protocol to enable on-demand streaming of real-time monitoring parameters, dynamically retrieved from a configurable set of network devices. The telemetry service is efficiently introduced for Software Defined Networking in EONs, also accounting for disaggregated network elements through standard YANG-defined network models. The implemented telemetry service is experimentally validated for partial and fully disaggregated architectures over networking scenarios where soft failures not addressable through traditional monitoring solutions are successfully detected and localized. Moreover, detailed performance evaluation is conducted varying the number of subscribed network elements, the type of communication (e.g., compressed vs. uncompressed, bundled vs. unbundled), showing remarkable scalability performance particularly in the case of compressed and bundled telemetry configurations.

A Tutorial on Machine Learning for Failure Management in Optical Networks

“…As of today, YANG models are moderately mature and preliminarily supported by several vendors, but they are not yet completely inter-operable, and this is significantly delaying their deployment in production networks. Second, a streaming telemetry protocol is required to efficiently retrieve data directly from devices to the Telemetry Collector running ML algorithms [71]. Moreover, telemetry should enable efficient data encoding, secure communication, subscription to desired data based on YANG models, and event-driven streaming activation/deactivation.…”

Monitoring and Data Analytics for Optical Networking: Benefits, Architectures, and Use Cases

“…For a brief description refer to Table 2. With these protocols available, the selection of the collection period is no longer limited to 15 minutes, and it can be reduced to, for example, one second [9]. Note that the shorter the collection period, the shorter the event that can be detected, as well as the shorter the time to detect degradations.…”

“…There are multiple ongoing standardization efforts within several technology areas, where most of the proposals are based on three main principles: data modeling language that provides structured data models for technology and function specific data points; management protocol for encoding and carrying the data model information; and the operational process governing how the protocol interface is used and connections are managed. In addition to research activity working on network telemetry (see, e.g., [9]), practical industrial projects exist, including: OpenConfig (see openconfig.net) and the Open-ROADM (see openroadm.org) efforts.…”

Experiment-based detection of service disruption attacks in optical networks using data analytics and unsupervised learning

“…8 The advances in coherent receivers with digital signal processing (DSP) capabilities provide an extensive OPM dataset for each received channel at their destination, and enable high-performance monitoring with short monitoring cycles. 9,10 However, this generates a large amount of data which needs to be processed by the SDON controller. For this reason, several works in the literature rely on the application of machine learning (ML) for monitoring and managing the performance of optical networks in a predictive and a reactive manner.…”