Orquestração dinâmica da telemetria de rede In-band em planos de dados programáveis

Abstract

In-band Network Telemetry (INT) is an emerging network monitoring paradigm. By collecting low-level telemetry items in near-real-time, INT can substantially increase visibility across the entire network - allowing, for example, the identification of short lived network behaviors such as micro-burst and other anomalies. Recent studies have focused on (i) developing INT mechanisms to increase network-wide visibility; and (ii) designing new monitoring solutions. However, little has been done to efficiently collect telemetry data from the network in this paradigm. This is particularly challenging as depending on which items of network telemetry are collected, this can degrade the visibility of the entire network in terms of consistency/freshness. In this Work we propose the Dynamic Probing Planning for In-band Network Telemetry (DyPro), as a heuristic solution capable of defining optimized polling cycles to collect in-band telemetry data in programmable network infrastructures. Our model, unlike other research efforts, aims to dynamically satisfy the requirements of monitoring applications in terms of INT data. Results show that DyPro can outperform high-end solutions by up to 5x in terms of percentage of satisfied monitoring applications.