Age-of-Information Aware In-band Network Telemetry for Better Network Predictability

Abstract

In-band network telemetry (INT) monitoring is getting traction in the recent years for its ability to offer high-granularity network information on per-hop, per-flow and end-to-end basis. Such detailed network monitoring helps in better network management and accurate network (re)conliguration, supporting automated network management and control. However, until now the collected INT information is used as is, without any classification based on its provided information value. In this paper we derive the Age of Information (AoI) mathematical model to calculate the INT hop-time average AoI and the impact of reporting on the INT AoI. Such metric can be used to reduce the network overhead in term of INT reporting by not sending old AoI INT and determining the information accuracy. We also show that the INT hop-time average AoI for a layer three INT implementation for a wireless network depends heavily on the communication latency between each hop and end hop, and that it is bounded by the end-to-end communication latency averaged over the number of hops.